Porphyry-related mineral deposits are giant geochemical anomalies in the Earth’s crust with orders-of-magnitude differences in the content and proportion of the three main ore metals Cu, Au, and Mo. Deposit formation a few kilometers below surface is the product of a chain of geologic processes operating at different scales in space and time. This paper explores each process in this chain with regard to optimizing the chances of forming these rare anomalies.

On the lithosphere scale, deposits with distinct metal ratios occur in provinces that formed during brief times of change in plate motions. Similar metal ratios of several deposits in such provinces compared with global rock reservoirs suggest preceding enrichment of Au or Mo in lithospheric regions giving rise to distinct ore provinces. The largest Cu-dominated deposits and provinces are traditionally explained by selective removal of Au during generation or subsequent evolution of mantle magmas, but the possibility of selective Cu pre-enrichment of lithosphere regions by long-term subduction cannot be dismissed, even though its mechanism remains speculative.

Evolution of hydrous basaltic melts to fertile magmas forming porphyry Cu deposits requires fractionation toward more H2O-rich magmas in the lower crust, as shown by their adakite-like trace element composition. The prevailing interpretation that this fractionation leads to significant loss of chalcophile ore metals by saturation and removal of magmatic sulfide might be inverted to a metal enrichment step, if the saturating sulfides are physically entrained with the melt fraction of rapidly ascending magmas.

Ascent of fertile magma delivers a large mass of H2O-rich ore fluid to the upper crust, along points of weakness in an overall compressive stress regime, within a limited duration as required by mass and heat balance constraints. Two mechanisms of rapid magma ascent are in debate: (1) wholesale emplacement of highly fractionated and volatile-rich granitic melt into a massive transcrustal channelway, from which fluids are exsolved by decompression starting in the lower crust, or (2) partly fractionated magmas filling a large upper crustal magma chamber, from which fluids are expelled by cooling and crystallization.

Transfer of ore-forming components to a hydrothermal ore fluid is optimized if the first saturating fluid is dense and Cl rich. This can be achieved by fluid saturation at high pressure, or after a moderately H2O rich intermediate-composition melt further crystallizes in an upper crustal reservoir before reaching fluid saturation. In either case, metals and S (needed for later hydrothermal sulfide precipitation) are transferred to the fluid together, no matter whether ore components are extracted from the silicate melt or liberated to the ore fluid by decomposition of magmatic sulfides.

Production and physical focusing of fluids in a crystallizing upper crustal magma chamber are controlled by the rate of heat loss to surrounding rocks. Fluid focusing, requiring large-scale lateral flow, spontaneously occurs in mushy magma because high water content and intermediate melt/crystal ratio support a network of interconnected tubes at the scale of mineral grains. Calculated cooling times of such fluid-producing magma reservoirs agree with the duration of hydrothermal ore formation measured by high-precision zircon geochronology, and both relate to the size of ore deposits.

Ore mineral precipitation requires controlled flow of S- and metal-rich fluids through a vein network, as shown by fluid inclusion studies. The degree of hydrothermal metal enrichment is optimized by the balance between fluid advection and the efficiency of cooling of the magmatic fluid plume by heat loss to convecting meteoric water. The depth of fluid production below surface controls the pressure-temperature (P-T) evolution along the upflow path of magmatic fluids. Different evolution paths controlling density, salinity, and phase state of fluids contribute to selective metal precipitation: porphyry Au deposits can form at shallow subvolcanic levels from extremely saline brine or salt melt; high-grade Au-Cu coprecipitation from coexisting and possibly rehomogenizing brine and vapor is most efficient at a depth of a few kilometers; whereas fluids cooling at greater depth tend to precipitate Cu ± Mo but transport Au selectively to shallower epithermal levels.

Exhumation and secondary oxidation and enrichment by groundwater finally determine the economics of a deposit, as well as the global potential of undiscovered metal resources available for future mining.

Porphyry-related magmatic-hydrothermal ore deposits are among the best-described metal resources and have been extensively reviewed in terms of their geology (Hedenquist and Lowenstern, 1994; Cooke et al., 2005; Seedorff et al., 2005; Sillitoe, 2010; Richards, 2011; Wilkinson, 2013; Dilles and John, 2020; Park et al., 2021) as well as their economic importance today and for the future (Kesler and Wilkinson, 2008; Singer et al., 2008; Sykes and Trench, 2014; Arndt et al., 2017). The overall process of ore formation is broadly known, thanks to research and exploration studies over the last decades. Seminal papers on geology and wall-rock alteration (e.g., Lowell and Guilbert, 1970; Sillitoe, 1973) and field-based timing relationships between magma emplacement and hydrothermal mineralization (Gustafson and Hunt, 1975; Dilles and Einaudi, 1992) remain an essential research base used by exploration practitioners today.

Large porphyry Cu deposits contain millions of tonnes of Cu in small rock volumes on the order of a cubic kilometer, enriched in concentration by two to three orders of magnitude compared to average rocks. Giant porphyry Cu deposits are extremely rare and anomalous features in the Earth’s crust, to the point that a small number of the largest deposits host most of the total Cu resource discovered so far (Singer et al., 2008). In this paper I develop the thesis that the formation of these exceptional geochemical anomalies is not the product of special triggers or of unusual complexity but the result of common geologic processes that are unusually well aligned in space and time. I use the image of a chain of processes in which all chain links are essential for successful ore formation (Fig. 1). Although contributing processes are common, local conditions may be unusual and parameters may be extreme to explain the rarity of giant porphyry deposits and their clustering in distinct provinces and time periods. I therefore explore each of the well-known processes along the path of a Cu atom, from the generation of magma in the Earth’s mantle to the precipitation of a grain of Cu sulfide in a vein, and ask the question “Which conditions optimize the chances to make a large and rich ore deposit?” Some of my arguments will be criticized as speculative, but the question about optimizing each process step must be valid, because each step in which Cu is dispersed rather than concentrated reduces the success rate of the following steps toward a giant ore deposit.

This contribution is based on four lines of investigation developed worldwide and at ETH Zürich over the last decades: analysis of the state and composition of mobile phases using fluid and melt inclusions, numerical modeling to link physical and chemical processes, geochronology for measuring the timescales of processes, and quantitative geologic field work using extensive industry data. I do not attempt to summarize observations on porphyry deposits, nor will I reproduce thermodynamic phase diagrams or discuss details of physical modeling published in earlier papers (see Appendix for supplementary detail). But I use this quantitative research to develop arguments based on principles of chemical thermodynamics, and heat and mass conservation, to interpret the dominant magmatic and hydrothermal processes that may explain common geologic observations.

Porphyry deposits show a large range of concentrations in three main ore elements Cu, Au, and Mo, to the point that some deposits are mined for a single commodity such as porphyry gold deposits (e.g., Kisladag; Baker et al., 2016) or porphyry-molybdenum deposits (e.g., Climax, Henderson; Seedorff and Einaudi, 2004). Metal ratios not only vary among deposits but also characterize large global provinces. In this section I argue that primary melts may be regionally enriched in Cu, Au, or Mo by preceding lithosphere-scale modification of their melt source, thus contributing to province-scale metal endowment.

Molybdenum provinces and their source magmas

Some of the largest Mo resources occur as by-products of normal porphyry Cu deposits (Seedorff et al., 2005), which have Cu/Mo mass ratios around 40 comparable to bulk continental crust (Rudnick and Gao, 2014) and ~300× lower than primitive mantle (McDonough and Sun, 1995). Climax-type Mo porphyries are highly enriched in Mo/Cu even compared with continental crust. Twelve of the 16 biggest Mo deposits occur in two regions—the Western USA and Northern China (Fig. 2), which are both Precambrian cratons assembled during the Paleoproterozoic (2.5–1.7 Ga) and incorporate even older Archean blocks. Low-degree remelting in the Phanerozoic led to intraplate granite magmatism (Farmer and De Paolo, 1984; Gao et al., 2018), and mineralizing granites reaching the upper crust are highly fractionated and commonly enriched in F, but radiogenic isotopic signatures indicate an influence of ancient deep lithosphere (White et al., 1981; Seedorff et al., 2005). The nature of this Mo-rich source material is debated: either Precambrian lower crust (Farmer and De Paolo, 1984) or mantle metasomatized by Paleoproterozoic subduction (Pettke et al., 2010; Gao et al., 2018). The potential role of a secular change in Mo geochemistry is open to speculation (App. I).

Province-scale Au/Cu variation of porphyry deposits

The Cu/Au ratio of deposits mined for Cu as their main commodity extends from an order of magnitude below that of Earth’s mantle to an order of magnitude above this dominant global reservoir (Fig. 3). The most valuable 14 deposits fall into two groups: eight giant Cu-dominant deposits (± Mo) contain low bulk Au/Cu, making Cu their main economic commodity, contrasting with six Cu-Au deposits with Au contributing comparably to their total value. Considering all deposits, however, the range in Au/Cu is continuous, so the grouping among the few richest deposits may be geologically less significant than the variation between Au-rich and Cu-dominant provinces.

Geochemical principles: Variably chalcophile metals in global reservoirs

Variations of Cu/Au in global reservoirs (Fig. 3, lines) started with early differentiation of the planet, which separated Earth’s composition (~chondritic meteorites) into a metallic core and the primitive mantle (McDonough and Sun, 1995; Salters and Stracke, 2004). Later modifications are dominated by the increasing tendency of Ag, Cu, Au, Pd, and Pt to partition into accessory iron sulfide liquid relative to silicate melt (Gannoun et al., 2016; Kiseeva et al., 2017). As a result, mid-ocean ridge basalts (MORB) have elevated Cu/Au and Cu concentrations (Salters and Stracke, 2004) but are low in Au (Pitcairn, 2011; Webber et al., 2013). In arc systems, mantle oxidation by subducted H2O and Fe+III can lead to complete consumption of sulfide (“mantle roasting”; Mungall, 2002), so that partial melting may take up all available Cu and Au, and resulting basalts have Cu/Au close to the local mantle wedge. The ratio between less chalcophile Pd to more chalcophile Pt can be used as a monitor of sulfide saturation (Cocker et al., 2015; Park et al., 2019) and indicates that at least some primitive arc basalts are initially sulfide undersaturated. Sulfide excess remaining in the mantle wedge, or subsequent sulfide removal from the basaltic melt, may increase or decrease Cu/Au in the silicate melt depending on whether the sulfide is liquid or solid (Li and Audétat, 2012; Li et al., 2021). In terms of absolute Cu concentrations, analyzed basalts and primitive silicate melt inclusions overlap in Cu content irrespective of tectonic setting (Lee et al., 2012; Chiaradia, 2014), but variation within a factor of at least 3× is permitted by the data (Cox et al., 2020; Grondahl and Zajacz, 2022).

For Au, there is clear evidence for selective addition to certain mantle domains related to the edge of former mantle plumes (yellow band in Fig. 3): basalts along the Mid-Atlantic Ridge increasing in Au content toward Iceland (Webber et al., 2013), anomalous Au enrichment in primitive arc magmas whose mantle source had been located at the edge of the Ontong Java Plateau (McInnes et al., 1999), and mantle xenoliths from Patagonia enriched in Au by mantle metasomatism at the periphery of the Karoo large igneous province (Tassara et al., 2018). Given that subcontinental lithospheric mantle is long-lived over billions of years and has experienced many plume events, it is likely that its metal ratios vary, including Au-rich regions (Griffin et al., 2009).

Giant Cu-dominant provinces: Marinate the mantle before roasting?

Explaining the world’s largest Cu-dominant deposits in distinct provinces, notably the central Andes, is a long-standing issue (Sillitoe, 1972; Mungall, 2002; Sillitoe and Perelló, 2005; Muñoz et al., 2012; Cocker et al., 2015; Kay, 2019; Park et al., 2021; Grondahl and Zajacz, 2022). What causes the first-order observation that their Cu/Au ratio is significantly higher than normal mantle (Fig. 3)? Most published suggestions explain the high Cu/Au by a deficit in Au, but none is entirely convincing considering the sheer size of this global Cu anomaly.

Hydrothermal ore formation in relatively deep porphyry deposits enhances Cu/Au by nonprecipitation of Au (Murakami et al., 2010), so that province-scale similarity in erosion level may have contributed to regional Cu predominance (Park et al., 2021). If the ore-forming fluids had provided Au and Cu in similar initial proportions (Ulrich et al., 1999), about 5× the quantity of Au contained in the Grasberg Cu-Au deposit would have been flushed through El Teniente alone. This seems questionable, given the rather shallow formation of this Cu deposit (Klemm et al., 2007; Spencer et al., 2015). Differences in fractionation degree of the fluid-producing magmas could cause the distinction between Cu-dominant provinces and Cu-Au provinces by depth-dependent H2O saturation. Park et al. (2019) plausibly suggested based on platinum group element (PGE) geochemistry that magmas fertile for porphyry ore formation saturate sulfide at a later stage of fractionation, just before fluid saturation, compared to barren magma systems losing chalcophile metals to early-saturating sulfides. Less plausibly, the authors imply that small differences in fractionation—e.g., at 75% instead of 80% crystallization—decide between Cu-dominant and Cu-Au deposits. This might explain variations in Au/Cu ratio between neighboring deposits at the same erosion level (e.g., Los Pelambres and Frontera; Perelló et al., 2012). But would all Chilean Cu-dominant deposits have become Cu-Au deposits if only the fluid had saturated at a slightly earlier stage in each fractionation history?

Selective retention of Au in the magma source or in deep cumulates is presently the best explanation of the Cu-dominant ore provinces (Richards, 2011; Cocker et al., 2015). It requires that Cu-poor but highly Au- and PGE-rich sulfide melt is removed from the parental magma, but the process is sensitive to sulfide abundance and phase state and also remains debated. Lee and Tang (2020) proposed that extreme fO2 due to garnet formation enhances Cu extraction by promoting (almost?) complete sulfide dissolution in the melt, but the redox effect of garnet was disputed (Holycross and Cottrell, 2023), and complete sulfide dissolution leaves the high Cu/Au ratio of the Chilean deposits unexplained (App. II).

Selective Cu enrichment in the mantle source by long-lasting subduction remains a tempting explanation for the unique Cu endowment of the central Andean province (Sillitoe, 1972). The original Cu content of fertile magmas is uncertain because those that actually produced these ores are never exposed in their pre-fluid-saturation state. The low Au/Cu ratio of Andean deposits is surprisingly similar to the metal proportions in oceanic crust (Fig. 3) subducted over 200 m.y. prior to ore formation. Could subducted Cu-bearing but Au-poor MORB, previously oxidized and hydrated by seawater interaction, generate a Cu-rich but Au-poor fluid in the long-lived subduction zone, thus increasing the Cu content of mantle sulfides in the frontal part of the overlying mantle wedge? Subsolidus Cu metasomatism of the mantle wedge would be favored by oxidizing slab fluids originating from a sediment-covered oceanic crust (Ague et al., 2022). If such “marinated” material was later transferred by subduction erosion (Rutland, 1971; Kay, 2019; Fig. 1) into the hotter melting zone of the mantle wedge, subsequent “roasting” (Mungall, 2002) might produce a primary melt with anomalously high Cu content but low Au/Cu ratio.

Water enrichment and magmatic sulfide saturation are two decisive lower to midcrustal processes contributing to the differentiation of primary mantle melts to ore-forming magmas. Fractional crystallization in the lower crust is affected by thickness and tectonic stress state of the lithosphere. This leads to diagnostic trace element compositions of upper crustal rocks associated with porphyry Cu ± Au ± Mo deposits.

Increasing magmatic H2O content by fractionation

Fractional crystallization of basaltic mantle melt with 1 to 3 wt % H2O to intermediate compositions with >4 wt % H2O is an essential requirement for subsequent formation of porphyry deposits of any Au/Cu (Rohrlach and Loucks, 2005; Chiaradia and Caricchi, 2017; Chiaradia, 2020). Copper and Au are incompatible with major silicates and oxides (Hsu et al., 2017; Iveson et al., 2018) so that ore metals are enriched in the evolving melt until accessory sulfides saturate. The fractionation toward fertile magmas starts at the base of the crust at high P, as shown by elevated bulk Al, Mg/Mg + Fe, and trace element concentrations reflecting plagioclase suppression but early removal of Al-rich amphibole ± garnet (Alonso-Perez et al., 2009). Mineralizing magmas invariably contain amphibole rather than pyroxene phenocrysts, indicating melt water contents >4 wt % (Naney, 1983). A hallmark of fertile magmas is their adakite-like trace element signature, with increasing Sr/Y because Sr-accommodating plagioclase is suppressed while Y is incorporated in high-P amphibole. This contrasts with normal calc-alkaline fractionation at lower P, which removes Sr-bearing plagioclase, whereas Y behaves incompatibly (Rohrlach and Loucks, 2005). Besides high Sr/Y, other trace element ratios including high V/Sc and La/Yb are indicators of high-P amphibole or garnet fractionation (Loucks, 2014). Bulk Zr concentration in fertile magmas is low because of Zr incorporation into Al-rich amphibole but also because of earlier zircon saturation in relatively cooler but H2O-rich magmas (Loucks and Fiorentini, 2023a). Zircon thereby attains characteristic trace element signatures including less negative Eu anomalies, making this resistant mineral a valuable pathfinder for fertile magmas even after alteration or weathering (Loucks et al., 2024). Adakite-like evolution is most prominent in magmas generating Cu-dominant deposits in areas of anomalously thick continental crust (Chiaradia, 2014), whereas less fractionated magmas forming Cu-Au deposits show a weaker adakite-like tendency (Loucks, 2012). Magmas at the Pebble Cu-Au-Mo deposit (Alaska) follow a normal calc-alkaline fractionation trend, indicating that fertile magma evolution is not restricted to deep and thick continental crust (Olson et al., 2017). I conclude that fractionation at mid to lower crustal P enriching H2O to >4 wt % is essential, but fractionation does not exclusively occur at the base of thickened crust to exceptionally high H2O content.

Compressive tectonic setting and contrasting durations of magmatic processes

Magma storage in the ductile regime above the mantle-crust boundary initiates fractionation toward H2O-rich fertile magmas (Annen et al., 2006). Horizontal compression favors accumulation of large sills of H2O-rich magmas with relatively low density by preventing immediate ascent through the crust (Loucks, 2021). A change from a long period of normal extensional arc volcanism to shorter periods of compression, stopping extrusive activity and initiating intrusive magma emplacement, is well documented from many porphyry provinces. Such changes can be caused by subduction reversals due to collision with ocean plateaus or older arcs (Solomon, 1990; Rohrlach and Loucks, 2005), transient slab flattening due to ridge subduction (Sillitoe, 1998; Sillitoe and Perelló, 2005; Gilmer et al., 2018), or continent collision (Richards, 2015; Moritz et al., 2016; Grosjean et al., 2022).

Geochronology of multiple intrusions in mineralized districts and in exposed lower crustal sections indicate that the duration of igneous activity eventually leading to fertile magmas is on the order of 10 m.y. (Rohrlach and Loucks, 2005; Lee et al., 2017; Rezeau et al., 2019; Lee et al., 2021; Nathwani et al., 2021; Stirling et al., 2023; Large et al., 2024). This is brief compared to preceding periods of subduction (plate-tectonic timescales of tens to a few hundred m.y.) but long compared to the timescale for formation of a porphyry copper orebody (~1 m.y. or less; below).

Magmatic sulfides: Lost on saturation or entrained with ascending melt?

Accessory sulfides occur in exposed lower crustal arc cumulates (Rezeau and Jagoutz, 2020) and in xenoliths transported by magmas in regions of hydrothermal ore deposits (Métrich et al., 1999; Kamenetsky et al., 2017; Chen et al., 2020; Du and Audétat, 2020). The Cu content of most arc volcanic rocks reaching the surface above thick-crust continental arcs decreases with fractionation, and they are deficient in Cu compared to similar magmas evolving in thin-crust oceanic arc settings (Chiaradia, 2014). The prevailing interpretation posits that magmas fractionating near the base of thick continental crust lose around two-thirds of the Cu initially available in basalts by physical removal of accessory sulfide to (ultra) mafic cumulates rich in amphibole ± garnet (App. II). Subsequent loss of these cumulates to the mantle leads to Cu depletion of thick-arc volcanic rocks and is also thought to explain the bulk deficit of Cu relative to less chalcophile Ag in the continental crust, compared to primitive mantle basalts (Lee et al., 2012). This interpretation is consistent with slight Cu enrichment in lower crustal arc cumulates (Chen et al., 2020; Rezeau and Jagoutz, 2020; Ahmad et al., 2021). However, as the richest porphyry Cu deposits are formed by evolved magmas above thick continental crust, this argument implies ore formation from the remaining Cu of magmas partly depleted in Cu (Chiaradia and Caricchi, 2017; Du and Audétat, 2020; Lee and Tang, 2020; Chelle-Michou and Rottier, 2021) and remains subject to conflicting interpretations (App. II).

A step of Cu loss in the process chain toward the richest Cu ore deposits could be avoided if sulfides are not physically removed to cumulates but partly entrained with the melt phase of the fractionating magma (Tomkins and Mavrogenes, 2003; Heinrich and Connolly, 2022), possibly assisted by bubble flotation in CO2-bearing magmas (Mungall et al., 2015). Entrainment of any suspended sulfide particles is favored by forceful ascent of magmas in the tectonic setting where porphyry copper deposits are found (Loucks, 2021), and the H2O-rich and oxidized composition of such fertile magmas promotes subsequent sulfide decomposition and transfer of sulfur and metals to ore-forming fluid in the upper crust (Lee and Tang, 2020; Heinrich and Connolly, 2022).

Magma ascent serves as an elevator for H2O and ore-forming components to the upper crust, where ore metals are deposited, primarily driven by the T difference between hot magmatic fluid and cooler country rocks. The fluid-producing magma reservoir not only needs to be large enough—evidently, larger magma bodies are capable of generating larger ore deposits—but later sections will show that it also needs to deliver the required fluid quantity within a limited duration of time.

Mass balance requirements, magma, and fluid volumes

The required minimal volume of primitive melt can be estimated, assuming optimal conditions of preceding melt generation, to form a giant porphyry deposit containing 100 Mt Cu. This is similar to El Teniente (Chile) as the largest known deposit dominated by one concentric ore shell (Cannell et al., 2005; Spencer et al., 2015). An assumed initial Cu concentration of 100 ppm in the basaltic melt is at the upper end of published estimates. Sourcing this Cu mass requires at least ~1012 t or ~360 km3 of basalt. We assume that this primitive magma contains 3% H2O (Chiaradia, 2020) and starts fractionating in the lower crust (App. III).

Figure 4 schematically depicts three scenarios of magma evolution and emplacement. Both scenarios for mineralization (A, C) are consistent with fractionation to adakite-like magmas reaching the upper crust (Loucks, 2021). Areas in the cartoons are scaled to volumes assuming cylindrical geometry. The cartoons distinguish between total solids (green cumulate bodies) and fractionated silicate melt just before ore fluid generation (pink). The melt is assumed to retain all H2O, Cu, and other necessary components provided by the original mantle magma (i.e., perfectly incompatible behavior of fluid components, irrespective of whether they stay dissolved in silicate melt or are entrained with magmatic sulfides). The volume of fluid generated at variable depths is shown by the blue pattern superimposed on the melt phase to illustrate that all H2O is eventually expelled, either due to melt decompression (“first boiling”) or due to crystallization of H2O-free minerals (“second boiling”; Burnham and Ohmoto, 1980). The fluid volume produced at each depth considers the P dependence of fluid density but ignores its smaller T dependence (Driesner, 2007).

Published interpretations of magma bodies generating porphyry deposits follow two contrasting scenarios, illustrated in Figure 4A and C. The resulting orebody is scaled to the red sphere for an ore grade of 1 wt % Cu. The intermediate scenario B may be closer to the normal state of hydrous magma systems in the continental crust.

The magma conduit hypothesis

Figure 4A assumes that fluid is produced by a highly hydrous melt that has been fractionated to granitic composition in the lower crust, as proposed originally by Rohrlach and Loucks (2005). For our simple assumptions, 360 km3 of basaltic magma with 3 wt % H2O separates into 270 km3 of dry solids and 90 km3 of residual melt with a 4× higher H2O content of 12 wt %. Fluid could be produced by crystallization of this volume of granitic melt in a large lower crustal sill, but this requires that the melt volume is cooled at depth. This is inhibited by the small T gradients in lower crustal “hot zones” (Annen et al., 2006), making highly fractionated melts in the lower crust effectively “immortal” (Loucks, 2021) and preventing selective extraction of fluids and their direct transfer to the site of ore formation. Efficient fluid production requires that the entire volume of highly hydrous melt is injected upward, as one or several batches, into a major transcrustal magma conduit. Fluid will saturate immediately across the entire depth extent, as a result of melt decompression during ascent, and exsolution is further promoted by crystallization and heat loss to gradually cooler country rocks. Near the base of the magma column (~21 km for the chosen assumptions; App. III), the produced fluid adds some 50% to the magma volume, increasing upward to a fluid volume exceeding the magma volume (>100%) at ~7 km. All the fluid ascends through the magma conduit and expands up to the site of hydrothermal Cu deposition to a total fluid volume exceeding that of the entire transcrustal magma conduit.

Direct magma conduits from depth may the simplest answer to why mineralized porphyries are invariably steep magmatic structures, in contrast to flat-roofed granites forming Sn-W veins and greisens by broadly distributed fluid exsolution (Fekete et al., 2016; Hong et al., 2017). Transcrustal conduits for superhydrous magmas have been proposed by Li et al. (2017), Gilmer et al. (2018), and Loucks (2021) based mainly on petrological and geochemical arguments (see also Urann et al., 2022). High H2O fugacity by fractionation to cool granitic melts near the base of thick continental is also inferred from trace element patterns in zircon (Loucks and Fiorentini, 2023b), but adakite-like patterns of magmas that fractionated in the lower crust may also be recorded by zircon forming later at lower P. The melt volume required for El Teniente scales to a transcrustal pipe with an average diameter of ~2.6 km, which is much greater than the combined cross-sectional area of mineralizing porphyries at El Teniente but comparable to the total footprint of hydrothermal ore (Spencer et al., 2015). The conduit scenario of Figure 4A faces mechanical challenges that have only recently been addressed (Abdullin et al., 2024). Extraction of the required fluid from the entire magma column requires extreme permeability for fluid flowing through simultaneously crystallizing and H2O-exsolving magma, approaching a fluidized suspension. High vertical connectivity is expected to transfer supralithostatic fluid pressure (Pfluid) through extended depth intervals, potentially leading to several times the load of overlying rocks at 5-km depth. High Pfluid gradients between the conduit and the surrounding crustal rocks are expected to lead to magma fragmentation, veining, and wall-rock alteration already at midcrustal levels. Geologic evidence for such processes is not described from exposed midcrustal arc sections (e.g., Otamendi et al., 2017; Klein and Jagoutz, 2021). As the overpressure and the volume of fluid must increase upward dramatically, I find it difficult to conceive how the ascent of this fluid + magma mixture could be stopped a few kilometers below the surface, as mapped in many porphyry deposits (Gustafson and Hunt, 1975; Dilles, 1987; Dilles and Einaudi, 1992; Seedorff and Einaudi, 2004; Redmond and Einaudi, 2010). This ascent mechanism seems more applicable to the emplacement of kimberlite pipes—indeed, showing deep fenitization of host rocks (Smith et al., 2004) and leading to diatreme eruption through the surface (Abersteiner et al., 2022).

The magma chamber hypothesis

The alternative scenario shown in Figure 4C assumes only partial fractionation of mantle magma in the lower crust to dioritic, monzonitic, or tonalitic composition with moderately elevated H2O content. Fifty percent crystallization would double H2O in the melt to 6 wt % and still lead to an adakite-like trace element signature (Loucks, 2021, table 3). This H2O-undersaturated melt can be injected through a dike to fill a large upper crustal magma chamber (Clemens and Mawer, 1992). Optimally, magmatic sulfides are either undersaturated due to high fO2 or entrained with rapidly rising melt, avoiding Cu loss and permitting 2× enrichment of Cu in the melt. Decompression initiates fluid saturation at ~200 MPa if the magma contains H2O as the only volatile (Baker and Alletti, 2012), but massive production of H2O-rich fluid only advances by cooling and crystallization in the upper crustal reservoir.

This scenario is commonly assumed for porphyry Cu systems (Sillitoe, 2010) and quantified for examples ranging from moderate to giant scale. Most famous and best exposed is the Yerington batholith, where partial magma fractionation at deeper crustal levels successively filled three intrusions of a nested upper crustal pluton. The last phase (Luhr Hill granite) was a magma chamber producing alternating pulses of porphyry dikes and vein-forming fluid, mapped at district to mine scale in tilted fault blocks (Proffett and Dilles, 1984; Dilles et al., 2015). Intrusion dimension is consistent with mass balance constraints regarding H2O and ore-forming components (Dilles, 1987; Cline and Bodnar, 1991). A similar scenario for the giant Bingham Canyon Cu-Au-Mo deposit explains combined evidence from geology, geophysics, and zircon petrochronology (below). At the Coroccohuayco and Quellaveco deposits (Peru), fractionation in the lower crust followed by fluid saturation from a large mid to upper crustal reservoir is indicated by the evolution of S and Cl concentrations in magmatic apatite and amphibole phenocrysts formed at different pressures (Chelle-Michou and Chiaradia, 2017; Nathwani et al., 2023).

The physical advantage of large upper crustal magma chambers for ore formation is their ability to deliver a large but controlled fluid flux, at a characteristic rate driven by the loss of heat to significantly cooler country rocks. This scenario necessitates rapid ascent of a moderately H2O rich magma to the upper crust (Chelle-Michou et al., 2014; Schöpa et al., 2017; Large et al., 2021; below) and is also consistent with geochronological data from exposed midcrustal plutonic sections (e.g., Klein et al., 2021). This geometry is partly inspired by large upper crustal magma bodies leading to caldera eruptions (Jellinek and DePaolo, 2003; Bachmann and Bergantz, 2004; Lipman et al., 2022). Rapid filling of a large reservoir with initially water-undersaturated magma followed by thermally limited crystallization and fluid generation helps avoid immediate eruption to the surface because magma chambers are mechanically stabilized at an optimal depth of ~6 ± 2 km by the balance between the mechanical strength of upper crustal country rocks, the buoyancy of magma, and the rates of replenishment and fluid production (Huber et al., 2019).

The transcrustal mush system of hydrous arc magmatism

Figure 4B illustrates the common scenario for the same input quantity of hydrous basalt undergoing vertically extended fractionation and melt transport between smaller intrusions. The flux of hydrous basalt from the underlying mantle can sustain such a transcrustal magma system over millions of years and deliver magma to brief events of active volcanism punctuating long periods of apparent dormancy (Cashman et al., 2017). Volcanic products typically contain a mixed cargo of antecrysts (Streck et al., 2005) originating from different levels indicated by mineral barometry (e.g., Krawczynski et al., 2012). Vertically extensive fractionation favors sequential filling of crustal plutons by smaller magma batches, shown by field evidence for mingling between variably fractionated magmas in composite intrusions (Leuthold et al., 2012). As a result of slow segregation, intrusive granitoids have less evolved bulk compositions than indicated by their mineral compositions—i.e., they represent cumulates of crystals precipitated from more fractionated melts that had percolated further in the mush column (Cornet et al., 2022). Examples of sequentially filled intrusions range from long-lived lower crustal complexes (Stirling et al., 2023) to midcrustal batholiths (Oberli et al., 2004; Samperton et al., 2015; Walker et al., 2015) to barren upper crustal plutons (Leuthold et al., 2012). The Adamello batholith (Italian Alps) has the required volume to form a giant ore deposit and formed at a depth similar to the magma chamber envisioned in scenario C (Nimis and Ulmer, 1998), but it was emplaced as multiple plutons over a period of at least 12 m.y. (Schaltegger et al., 2019). Even small plutons within it were filled slowly by numerous magma pulses, as shown by internal contacts and deeper antecrysts dated by high-precision geochronology (John and Blundy, 1993; Schoene et al., 2012). Such composite intrusions contrast with the Yerington batholith in which the mineralizing Luhr Hill intrusion was filled rapidly by melt from a region of mid to lower crustal fractionation (Dilles, 1987; Schöpa et al., 2017).

The Altiplano-Puna magmatic complex in the central Andes is the largest coherent mush body and contains the largest volume of hydrous silicate melt presently existing on Earth, with a horizontal diameter in excess of 100 km. It is the product of >40 m.y. of magmatism and contractional tectonics, which had thickened the continental crust to ~70 km to elevate the present Altiplano-Puna plateau (Sparks et al., 2008; Wörner et al., 2018). Geophysical data indicate continuous fluid saturation from ~20% melt in magma mush, from at least 30-km depth to the surface (Gottsmann et al., 2022). Its dimensions would suffice to make 10 to 100 Cu deposits of the size of El Teniente (cf. Pritchard and Gregg, 2016), but fluid focusing seems limited and epithermal deposits of appropriate dimension are unknown (Gorini et al., 2018). Proponents of ore formation by transcrustal conduits feeding fertile magmas toward the surface (Fig. 4A) may argue that the low mechanical strength of this giant mush body prevents throughgoing channelways for focused ascent of fertile magma from the lower crust. They may also point to the dominantly crustal-anatectic composition of lavas reaching the surface above the Altiplano-Puna magmatic complex (Godoy et al., 2019). Proponents of large upper crustal magma chambers as sources of ore fluid may additionally argue that the complex is too deep and thermally stable to allow rapid fluid generation and that its crystal fraction is too high for large-scale lateral collection of fluids from the crystal mush. Such mush systems are able to produce large melt-dominated magma chambers (Bachmann and Huber, 2016), but these melts do not generally show the geochemical signature of plagioclase-suppressed, lower crustal fractionation (Loucks, 2014). Opinions differ (e.g., Hudson et al., 2022), but compared to either of the two scenarios of Figure 4A and C, transcrustal mush systems seem less likely to produce the large, rapid, and focused fluid ascent needed for ore formation.

The transcrustal mush state of magmatic arcs (Fig. 4B) is nevertheless important, because it typically precedes and prepares the crust for scenario C. Thermal preparation of the upper crust by extended magmatism allows the emplacement and characteristic lifetime of large upper crustal magma chambers. Heating of the crust over millions of years (Karakas et al., 2017) creates relatively ductile regions in the otherwise brittle crust, allowing emplacement of large hydrous but nonerupting magma chambers at 4- to 8-km depth (Huber et al., 2019). Local plasticity in the upper crust is shown by intrusion-related folding around large plutons (Brack, 1981; Brun et al., 1990; Marko and Yoshinobu, 2011). Thermal preparation of the crust is reflected by multimillion-year magmatism in the lead-up to one or several brief events of mineralization in most major porphyry systems (Maughan et al., 2002; Halter et al., 2004; Rohrlach and Loucks, 2005; Sillitoe, 2010; Lee et al., 2017; Nathwani et al., 2021; Large et al., 2024). Transtensional jog openings in compressional tectonic settings can localize forceful ascent of fertile magmas and ore-forming volatiles into preheated upper crust (Tosdal and Dilles, 2020).

Efficient transfer of S, Cl, and ore metals to the exsolving magmatic-hydrothermal fluid leads to one to two orders of chemical enrichment, simply because the fluid represents only a few percent of the melt yet takes up a large portion of these components (Candela and Holland, 1984). Generation of an effective ore fluid faces two requirements. First, it should minimize loss of ore-forming components to minerals locking them away into magmatic rock. Second, the transit from melt to hydrothermal fluid should not separate ore-forming components that are later required together for mineral precipitation—notably ore metals from sulfur needed for sulfide deposition.

Principles: Fluid exsolution and trace element extraction from crystallizing magma

The transfer of ore metals from magma to an exsolving hydrothermal fluid is controlled by chemical competition among crystals, melt, and fluid. Therefore, the concentrations in melt and fluid are related to the extent of crystallization. For example, a melt crystallizing to the point F in Figure 5 becomes enriched in H2O to 5%, where it reaches fluid saturation. From this point onward, the H2O fugacity (~concentration in the melt) stays constant at constant pressure (Pfluid = P), and H2O is expelled as the magma crystallizes to complete solidification (from F to 100% crystals on lower axis, from 0 to 100% H2O expelled in top axis).

The behavior of a trace metal depends on its tendency to transfer to the fluid phase in comparison to that of H2O. This is illustrated by three colors in Figure 5A (concentrations in the fluid, with bubble size showing the fraction of each element transferred to the fluid in each crystallization step) and Figure 5B (concentration of the respective elements in the remaining melt). Three different distribution coefficients Di = Ci(fluid)/Ci(melt) are assumed. The blue trace metal prefers the fluid phase with a distribution coefficient Dblue = 50, which is greater than CH2O(fluid)/CH2O(melt) ≈ 100%/5% = 20, so that it will become enriched in the fluid phase but depleted in the remaining melt; we might call this an “enthusiastically fluid-loving” element. By contrast, for the red element Dred = 5 is positive but smaller than CH2O(fluid)/CH2O(melt), so its concentration in the fluid is still greater than in the melt, but increasing crystallization raises the concentrations in the remaining melt and in the fluid; call the red element “reluctantly fluid-loving.” If the distribution coefficient Dgray equals CH2O(fluid)/CH2O(melt), the gray element and the main fluid component H2O are expelled in the same proportions; hence, concentrations in fluid and in remaining melt stay constant (Audétat, 2019). The fraction of a trace element that is transferred to the fluid in each step of fluid exsolution is greatest at the beginning for an enthusiastically fluid-loving element (blue), delayed toward the end for a reluctantly fluid-loving element (red), or extracted in equal steps in the neutral case (gray).

This concept of open-system fractionation or Rayleigh distillation has been applied to coexisting melt and fluid inclusions from late-magmatic veins and miarolitic cavities analyzed by laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS), yielding field-based partition coefficients for numerous elements (Zajacz et al., 2008; Audétat, 2019). Cesium is incompatible even with late-crystallizing minerals, but its partition coefficient to the fluid is only a little above 1, so its enrichment is a proxy for the degree of magma crystallization. Plotting element concentrations of fluid inclusions vs. their Cs concentration measures the degree of preference of various trace metals between melt and fluid. With this approach, Audétat (2019 and earlier papers) showed that most ore-forming elements including Cu, Pb, Mo, and Cl itself are enthusiastically fluid-loving at high P (~200 MPa or higher) but only reluctantly fluid-loving at ~100 MPa or lower. This P dependence was first documented experimentally for NaCl by Kilinc and Burnham (1972), which served as a basis for the pioneering modeling of ore fluid generation by Cline and Bodnar (1991).

Material properties: Distribution coefficients vary and depend on each other

More recent experimental work has shown that distribution coefficients (D) between silicate melt and fluid are not constant, as assumed for the simple illustration of Figure 5, but vary with P and composition owing to changing thermodynamic properties of melt, fluid, and dissolved complexes.

Chlorine: Cl is a decisive component for ore formation, and its melt/fluid distribution coefficient is affected by several factors:

  1. Solvation by H2O molecules (hydration) stabilizes Cl and other salt-like components in the fluid and strongly increases from gas-like fluids (minimal hydration) to liquid-like aqueous fluids (Seward et al., 2014).

  2. NaCl (encompassing all chloride species in the fluid, except for HCl dominating in volcanic gases) makes fluid denser and more liquid-like, and this nonideal mixing causes phase separation into a low-density H2O-rich vapor phase and a denser NaCl-rich liquid below ~150 MPa (Sourirajan and Kennedy, 1962; Driesner and Heinrich, 2007).

  3. The stability of Cl itself is enhanced by increasing Cl concentrations in the fluid.

  4. At the same time, Cl in the silicate melt is destabilized by increasing SiO2 concentration in the melt due to polymerization.

The four trends together cause DCl(fluid/melt) to increase steeply with increasing P, fluid salinity, and SiO2 concentration in fractionating silicate melt (Tattitch et al., 2021).

Copper transfer: Cu transfer from melt to fluid depends on fluid density (hence P) as well as redox conditions (Zajacz et al., 2012). It largely follows the behavior of Cl owing to the formation of strong complexes like CuCl(aq)0 and CuCl2(aq) at higher salinities, as well as mixed Cl-S complexes in H2S-rich fluid systems (Hemley et al., 1992; Liu and McPhail, 2005; Audétat and Simon, 2012; Tattitch and Blundy, 2017). Similar relationships apply to other Cl-complexed ore metals including Sn, Pb, Zn, and Mn (Iveson et al., 2019). FeCl2 shows comparable behavior but saturates as igneous magnetite and is a major component of magmatic fluids besides NaCl and KCl. Even trace metals with oxide or sulfide coordination in aqueous fluid prefer the fluid phase with increasing salinity (Audétat, 2019) because of the formation of polymetallic complexes with alkali metals such as Mo(Na,K)HO4(aq) (Tattitch and Blundy, 2017; Fang and Audétat, 2022) or Au-Na-H-S species that may be particularly important in mafic alkaline magmas (Zajacz et al., 2010).

Sulfur: S has a strong preference for the fluid as molecular SO2 and H2S species, depending less on hydration, so S is readily transferred to the fluid at all pressures (Seo et al., 2009; Masotta et al., 2016). Sulfur distribution between crystallizing magma and fluid is sensitive to redox potential because volatile S species compete with sulfate (SO4) as stable components of silicate melt, leading to anhydrite saturation in oxidized magmas associated with porphyry Cu mineralization (Luhr et al., 1984; Carroll and Rutherford, 1985; Audétat et al., 2004; Hutchinson and Dilles, 2019; Kleinsasser et al., 2022).

Stepwise fluid extraction: Early vs. late metal transfer

To illustrate the influence of depth and magma composition on metal extraction, four distillation models were calculated from the point of fluid saturation to 100% solidification. The calculations extend those of Candela (1989), Cline and Bodnar (1991), and Audétat (2019) and consider variable DCu, DMo, and DCl following equations and experimental data of Tattitch and Blundy (2017) and Tattitch et al. (2021). Figure 6B to D is based on 100 ppm Cu and 2 ppm Mo in an initial mantle melt that had been fractionated to variable degrees prior to fluid saturation, extending the scenarios of Figure 4. Fluid saturation occurs at different P determined by the H2O content of the melt. Increasing SiO2 content in the melt raises DCl, which in turn depends on fluid salinity (Tattitch et al., 2021). Scaling the fractions of transferred components by symbol size focuses on the importance of early versus late stages of metal extraction in the following four cases, discussed from deep (Fig. 6D) to shallow (Fig. 6A; see App. IV for calculations).

High degrees of fractionation to a melt with 70 wt % SiO2 and 12 wt % H2O saturates an aqueous fluid in the lower crust (Fig. 6D), approximating the magma-conduit scenario previously explored in Figure 4A. At the start of fluid extraction, the highly fractionated melt is ~4× enriched in H2O, Cl, and ore metal contents compared to the primitive basalt. High SiO2 in the melt and high fluid density lead to a high DCl for the first-saturating fluid batches. DCu is even more elevated, so that 73% of the available Cl and 95% of the Cu are transferred to the first quarter of exsolving fluid. Metal enrichment in early-exsolving fluid is favorable for later ore formation, because volatile sulfur species (H2S and SO2) also exsolve into early-saturating fluids, particularly in silicic and oxidized magmas where SO2 predominates in the fluid (Audétat and Simon, 2012; Masotta et al., 2016). Sulfur is therefore available together with CuCl and FeCl2 for later precipitation of Cu-Fe sulfides in the upper crust. This point was initially made by Rohrlach and Loucks (2005) and makes fluid production at high P after advanced lower crustal fractionation an attractive process for porphyry copper mineralization.

Fractionating a basaltic magma with the same initial quantities of H2O, Cl, Cu, and Mo to intermediate composition (~62 wt %SiO2; Fig. 6C) will enrich incompatible volatiles in the melt to only ~2× initial concentration. The lower H2O concentration delays fluid saturation until this magma reaches 200 MPa in the upper crust. Lower P, and lower SiO2 and Cl in the melt, lead to lower DCl and DCu. After saturation of an initial fluid low in Cl and Cu, these components accumulate in the remaining melt until SiO2 is sufficiently high to eventually transfer Cl, Cu, and also Mo into late-saturating fluid batches. This behavior is less favorable for ore formation for several reasons. First, Cu is spread over many fluid batches, leading to overall lower Cu concentrations in the fluid. Second, ore metal liberation is delayed relative to the expulsion of sulfur, making it less likely that all essential components for sulfide precipitation later meet at the ore deposition site. And third, the production of the most Cu-rich fluid from already largely crystallized magma (>60%) inhibits focused extraction of this fluid (Parmigiani et al., 2011) and tends to leave the metals dispersed in plutonic rocks.

This unfavorable situation can be inverted by increasing the Cl/H2O in the parental magma, as shown in Figure 6B for the same initial Cl, Cu, and Mo concentrations but only half the initial H2O content. This creates conditions in the upper crust that are similar to or even better than the lower crustal scenario D. Somewhat lower H2O in the primary basalt allows a moderately fractionated, intermediate-composition melt to be injected into an upper crustal magma reservoir without immediate fluid saturation, where it starts crystallizing at an assumed 200 MPa (same P as Fig. 6C but in a geologic scenario corresponding to Fig. 4C). Fluid will only saturate after approximately 30% crystals have formed, by which stage the interstitial melt has been enriched to ~70% SiO2. The higher Cl/H2O ratio and SiO2 content of this residual melt elevates DCl sufficiently to cause Cl, and to even greater degree Cu, transfer to the first batches of exsolving fluid, together with any volatile sulfur species; 84% of available Cu is extracted into the first quarter of exsolved fluid. At this point, the overall crystallinity of the magma (including the crystals formed prior to fluid saturation) has reached ~45%. This is within the 40 to 70% crystallinity window, where physical extraction of fluid is particularly efficient (Parmigiani et al., 2011; below). The later batches of fluid are largely exhausted of Cu but still contain an appreciable fraction of the Mo available in the system. The evolution of Figures 4C and 6B matches interpretations of fluid and melt inclusions from the Bingham Canyon deposit (Landtwing et al., 2010; Seo et al., 2012; Grondahl and Zajacz, 2017) as well as petrographic, geophysical, and zircon-geochemical data discussed later.

Figure 6A illustrates very shallow fluid exsolution at ~100 MPa, where fluid density has a dominant influence on distribution coefficients. Low-density vapor is a poor solvent for salt and metals so that they stay in the melt phase while the first batches of single-phase vapor are extracted. Chlorine accumulates in the melt until its concentration is high enough to separate a highly saline liquid coexisting with low-salinity vapor. This phase separation is not modeled here, but extrapolated average salinities are shown by stippled symbols. Granite fractionation to such high Cl and metal but low H2O content is unlikely in arc systems without earlier H2O and metal loss at greater depth. Locally, this behavior is shown by fluid inclusions from miarolitic cavities in shallow granitoid intrusions (Audétat, 2019). The model also matches P-T estimates and percent-level Cu concentrations in extremely saline brine inclusions or hydrous salt melts in quartz fragments ejected by the shallow caldera-forming magma chamber of Cerro Escorial (Chile) after loss of H2O and volatile S (Fiedrich et al., 2020). Indeed, no ejecta indicating ore formation at depth were discovered here, despite intensive field searching.

The purpose of these model calculations is to show the impact of magma fractionation and depth as the two leading factors allowing effective ore fluid generation. The examples in Figure 6 show trends rather than quantitative predictions, firstly because of considerable experimental uncertainties. Secondly, fluids are not expelled from large magma reservoirs in separate batches as modeled, but they partly remix from regions of variable composition and crystallinity before reaching an ore deposit (Chelle-Michou et al., 2017). As a result, fluids in porphyry deposits are more constant in composition and lower in metal concentration than the extremes in Figure 6. Nevertheless, metal transfer together with volatile S into to early-saturating fluid is favorable for ore formation, no matter whether the ore components are extracted from silicate melt or by decomposition of magmatic sulfides (Hattori and Keith, 2001; Halter et al., 2005; Chelle-Michou and Rottier, 2021; see App. IV for further discussion).

Porphyry Cu ore is hosted by vein networks associated with small intrusions with a high proportion of phenocrysts (typically 50 vol %; Seedorff et al., 2005) originating from a larger volume of fluid-producing magma at depth. This section explores the physics of fluid extraction from this magma and concludes that cooling of a large upper crustal magma chamber can lead to spontaneous focusing of ore fluids (Candela, 1991). The proposed model process avoids catastrophic eruption and occurs at timescales in accord with high-precision geochronology.

Rapid generation and self-focusing of magmatic fluids

Fluid flow through magma is linked with magma crystallization, which drives the generation of fluid by excluding H2O and other volatiles from crystals, but these processes also change melt viscosity and effective permeability of a cooling magma. Such complex feedbacks require numerical simulation to link large-scale fluid flow to the grain-scale interaction between crystals, fluid, and melt. The focus here is on exploring conditions for ore fluid generation; other published approaches to modeling emphasize the even shorter times of vein formation (Cathles and Shannon, 2007) or different aspects of magma reservoirs and hydrothermal fluid flow, combining different subsets of similar basic equations (e.g., Ingebritsen and Appold, 2012; Sparks et al., 2019).

Mechanical processes in a vertically extensive magma conduit have been considered for Climax-type porphyry Mo deposits, assuming that the H2O content of melt is low enough to saturate fluid only in the upper part of the magma column (Shinohara et al., 1995). The assumed high T and low viscosity of these F-rich granitic melts allows vertical convection within the magma conduit. Convection is driven by the density reduction caused by suspended bubbles, feeding a steady supply of magmatic fluids to the apex of the magma column, and by the increased density of H2O-depleted melt. A similar mechanism was proposed for porphyry Cu-Au deposits formed by rather H2O-poor dioritic magmas (Cloos, 2001). The mechanical behavior of a super-wet granitic melt saturating fluid already in the lower crust (Fig. 4A) has first been addressed by Abdullin et al. (2024), whereas processes in an upper crustal magma chamber similar to Figure 4C were modeled in some detail.

Figure 7 combines results from three partly coupled simulations of an upper crustal magma chamber of comparable size to that in Figure 4C (Weis et al., 2012; Andersen and Weis, 2020; Lamy-Chappuis et al., 2020; App. V). It incorporates grain-scale magma behavior simulated with particle-based numerical methods, which show that simultaneous crystallization and fluid exsolution defines a window of crystallinity and water production where fluid escapes efficiently from the remaining melt, because crystals touch each other and support a contiguous tube network (Parmigiani et al., 2011). This microscopic behavior has been translated into effective permeability of crystal mush and incorporated into a finite element model for fluid flow within a cooling and crystallizing magma chamber (Lamy-Chappuis et al., 2020). Numerical results (Fig. 7) predict the development of a zone of high permeability at a crystal fraction between 40 and 70%, along which fluids are produced by crystallization and flow laterally through the tube network to the point of lowest P. This tube-flow zone advances inward at the expense of a more melt-rich and therefore less permeable interior of the magma chamber. To the outer side, the tube-flow zone is sealed by crystal mush in which the high crystal fraction requires forceful expulsion for fluid to escape. The surrounding carapace of hot igneous rock can only be transgressed by hydrofracturing when Pfluid exceeds lithostatic pressure plus rock strength. These inward-retracting zones produce and confine fluids, thereby focusing the available volatiles to the same point during cooling of the intrusion. Focusing is directed to the point of lowest confining pressure, which can be modified by preexisting structures that are also likely to determine the location of magma ascent through the underlying crust. The resulting hot fluid plume and its interaction with cooler groundwater-saturated host rocks was first described by Henley and McNabb (1978), and its physical hydrology was evaluated by Weis et al. (2012), including variation of rock permeability in response to changing T and P (Fournier, 1999; Ingebritsen and Manning, 2010). A stable but internally fluctuating fluid plume develops from feedbacks between fluid overpressure (leading to vein opening), alternating with heat advection by the fluid and quartz precipitation (leading to plastic closure of permeability). These competing effects on permeability stabilize a front of sharply decreasing P and T over a small flow distance. As the solubilities of quartz decreases with P, whereas Cu-Fe sulfides solubility decreases steeply with falling T, the model predicts a column of quartz precipitation and a more confined ore shell of Cu-Fe sulfide deposition due to the steep P-T gradient near the outer front of magmatic fluids against convecting external fluids, which remove heat and transport it toward the surface (Cathles, 1977). Depending on depth of the magma chamber, fluid in the region of quartz precipitation separates into brine and vapor. The vapor may ascend at a greater velocity than the denser and more viscous brine and thereby contract to an H2O-rich liquid, becoming miscible again with the brine or with liquid meteoric water (below).

This model explains the lateral flow required to collect fluid from a much larger magma volume losing heat to the surface. At the deposit scale, it explains hydrothermal mineralogy by showing that porphyry ore shells typically form in the upper and outer zones of a column of repeated quartz veining and intense potassic alteration (Proffett, 2003; Landtwing et al., 2010; Spencer et al., 2015; Reed and Dilles, 2020). Fluid inclusions show that these vein stockworks represent former temperature anomalies compared to surrounding cooler country rocks, in which propylitic alteration halos are much wider than expected from contact metamorphism by porphyry intrusions (e.g., Hezarkhani and Williams-Jones, 1998; Wilkinson et al., 2020). Both observations are explained by the dominant heat advection by focused magmatic fluid and the heat loss to external rocks saturated with convecting meteoric water. The hydrologic interface can be quite sharp and only locally involves some fluid mixing (Eastoe, 1978; Henley and McNabb, 1978; Fekete et al., 2016). Stabilizing the site of Cu deposition by the competing P-T effect of permeability creation and its closure is essential for accumulating economic ore grades, rather than dispersing metals along the upflow path or across a larger horizontal area (Fournier, 1999; Heinrich et al., 2005). The point of mineralization is commonly guided by structural weaknesses, but fluctuating fluid overpressure is the main driver for the crisscrossing, dominantly vertical and radial orientation of quartz veins (Titley and Heidrick, 1978; Gruen et al., 2010). External stress becomes more important at lower T and near-hydrostatic Pfluid, where faults dominantly localize epithermal veins (Tosdal and Dilles, 2020; Codeço et al., 2022). A magmatic fluid plume, creating a P-T anomaly compared to the surrounding hydrothermal cooling engine, requires that fluid expulsion is sufficiently rapid to prevent cooling and low-T alteration of rocks between events of porphyry emplacement, which is not generally observed in detailed field studies (Dilles and Einaudi, 1992; Proffett, 2003; Redmond and Einaudi, 2010).

Duration of mineralization: Longer for bigger tonnage; as fast as possible for best ore grade

The timescale of fluid flow and ore formation predicted by multiphysics modeling is consistent with high-precision geochronology, indicating a correlation between deposit size and duration of mineralization (Chelle-Michou et al., 2017). Some early geochronological studies have emphasized the multimillion-year duration of ore-related magmatism (e.g., Sillitoe and Mortensen, 2010; Simmons et al., 2013), which was sometimes mistaken to imply that extended multipulse mineralization is itself beneficial to forming rich ore deposits. The interpretation of geochronological data has been sharpened with the advent of chemical abrasion-isotope dilution-thermal ionization mass spectrometry (CA-ID-TIMS) of single zircon crystals, which can resolve the times of zircon crystallization within the thermal history of magmatic rocks (Miller et al., 2007). Linking high-precision geochronology with chemical characterization of zircon grains by LA-ICP-MS microanalysis relates time to chemical evolution of magmas (“petrochronology” pioneered by Chelle-Michou et al., 2014; Samperton et al., 2015).

Modern petrochronology has been applied to several porphyry ore systems where field mapping has documented periods of hydrothermal vein formation bracketed by events of magma emplacement (Fig. 8). Such outcrop relations allow definition of two types of “durations” or “lifetimes” in magmatic-hydrothermal ore systems (von Quadt et al., 2011). On one hand, the total variation of precise and concordant ages of all zircon grains from a suite of porphyry samples defines the total lifetime of the magma reservoir from zircon saturation through ore fluid generation, including possible termination by catastrophic volcanic eruption (Buret et al., 2017). On the other hand, emplacement ages of porphyries bracket the duration of hydrothermal vein formation. If a sufficiently large number of zircon phenocrysts is dated, the youngest zircons approximate the termination of zircon crystallization in a particular magma batch and define the age of porphyry emplacement (e.g., Large et al., 2020; see App. V for explanation and further references).

Figure 8A summarizes results from six deposits of variable Cu tonnage for which single-crystal zircon ages were resolved based on clear field relationships among intrusion samples. Results are based on CA-ID-TIMS measurements, except for El Teniente where LA-ICP-MS data suffice for resolving the age brackets of interest because of the young age and long duration of mineralization at this deposit. Each cited study discusses specific assumptions and ambiguities, including the interpretation of the few oldest and youngest zircon grain ages, but all allow distinction between the longer lifetime of mineralizing magma systems (gray bars) and a shorter duration of hydrothermal mineralization bracketed by pre/syn- and post-ore porphyries (yellow bars). The top of the gray bars is invariably predated by even older (5–2 Ma) intrusions in the same district showing no evidence of ore fluid production and commonly distinct trace element characteristics. Inset 8B explains the interpretation approach to magma lifetime and mineralization duration, with their analytical uncertainties shown as fainter extensions (Large et al., 2020). At Bingham Canyon, a broad cupola of fine-grained equigranular monzonite (EM) is intruded by two pre/syn-mineralization porphyry dikes (QMP, LP) followed by quartz veining and Cu introduction with decreasing intensity, as indicated by a waning color gradient in Figure 8. Most Cu-Au mineralization was completed prior to the latest but still comagmatic intrusion phase (QLP) that introduced little Cu but most of the Mo (Fig. 8C). Copper and Mo ore shells are concentric, with Mo displaced slightly downward (Porter et al., 2012), indicating that the thermal anomaly of the focused fluid plume (Fig. 7) was maintained over more than 600,000 years without intervening low-T (<350°C) alteration and veining (Redmond and Einaudi, 2010).

The combination of physical modeling (Fig. 7) and high-resolution petrochronology (Fig. 8) explains why the duration of mineralization correlates with the Cu tonnage of porphyry deposits: larger magma reservoirs can deliver more ore fluid but take longer to cool and expel their fluid. Ore grade, as the other factor determining economic viability, is primarily determined by the efficiency of cooling a focused fluid plume by heat loss to convecting surface water, located by structure and host-rock permeability. High ore grade requires flux rates enabling a sharp metal solubility gradient at a hydrologically stabilized precipitation front to accumulate Cu within a restricted ore volume.

Physical processes control the chemistry of mineral precipitation as a major step of hydrothermal metal enrichment. Geologic, mineralogical, and analytical observations at mine to microscope scale can be linked to experimental data defining material properties of fluids and minerals. This section argues that the initial chemical composition of magmatic fluid, together with changes of fluid properties with decreasing P and T, dominate the efficiency of selective ore metal deposition.

Linking geologic observations to laboratory experiments

Veins and hydrothermal alteration: Multiple intersecting vein and overprinting wall-rock alteration types in porphyry deposits seem bewildering, but show commonalities and systematic variations in space and time. Intersection relationships with successive events of magma emplacement provide timelines that can be related to deposit-scale zoning of mineralogy, ore grade, and metal ratio. Vein types show systematic relationships to ore composition, alteration type, and the distribution of ore composition relative to the paleo-surface, as deduced by deposit-scale geologic mapping (Gustafson and Hunt, 1975; Hedenquist et al., 1998; Proffett, 2003; Seedorff and Einaudi, 2004; Setyandhaka et al., 2008; Redmond and Einaudi, 2010; Reed et al., 2013; Spencer et al., 2015). Figure 9 shows a few globally widespread vein and alteration types, sorted by increasing inferred formation depth. Economic Cu grade is broadly correlated with the density of quartz veining, which is associated with potassic alteration (K-feldspar and hydrothermal biotite replacing magmatic minerals by exchanging Ca ± Na for K in the fluid; e.g., Proffett, 2009) or, especially in relatively deep deposits, alteration halos of muscovite + biotite ± andalusite with disseminated chalcopyrite and bornite (Perelló et al., 2012; Reed and Dilles, 2020). On the other hand, microscale study of quartz vein textures by scanning electron microscopy-cathodoluminescence (SEM-CL) shows that Cu-Fe sulfides rarely occur as inclusions in the main generation of vein quartz but are almost invariably in contact with microveinlets and dissolved grain boundaries filled with a second generation of typically dull-luminescent quartz (Rusk and Reed, 2002; Redmond et al., 2004; Schirra et al., 2022; Fig. 9C, inset). These observations have led to conflicting interpretations. One interpretation is that Cu-Fe sulfides precipitated early with quartz veining and alteration in the biotite stability field and were later remobilized into microcracks with dull-luminescing quartz (Cernuschi et al., 2023). Alternatively, bornite + chalcopyrite were introduced after the main quartz veins, aided by secondary permeability in the window of retrograde quartz solubility below 450°C (Landtwing et al., 2010; Stefanova et al., 2014; Monecke et al., 2018; Jensen et al., 2022) and in an overprinting sulfide-rich but quartz-free generation of cracks (“paint veins”) postdating the dominant vein quartz and associated with incipient feldspar-destructive muscovite ± chlorite alteration (Hedenquist et al., 1998; Klemm et al., 2007; Schirra et al., 2022; Fig. 9C). Finally, all quartz and Cu-Fe-sulfide–rich veins are overprinted by feldspar-destructive alteration (muscovite and clay minerals), which is pervasive in the upper parts of many porphyry deposits and extends along pyrite-filled fractures to greater depth (Gustafson and Hunt, 1975), channeling low-salinity magmatic fluids to overlying epithermal deposits (Hedenquist et al., 1998).

Fluid inclusions: Texturally coeval groups of inclusions with similar phase proportions represent samples of a previously homogeneous fluid (Roedder, 1984; Goldstein and Reynolds, 1994). For example, a 70 vol % vapor bubble plus 30 vol % aqueous liquid seen in a fluid inclusion at ambient T shows that the original fluid trapped at high T had a bulk density near 0.3 g/cm3. The size of halite daughter crystals is a measure of bulk salinity in excess of ~26 wt % NaCl equiv, the solubility of halite at ambient T. Petrography allows definition of the time sequence between inclusion entrapment and stages of mineral precipitation (assisted by SEM-CL imaging of vein textures; Rusk and Reed, 2002; Fig. 9C, inset). Fluid inclusions show that weakly saline magmatic fluids entering the base of porphyry Cu deposits separate into highly saline liquid and lower-density salt-poor vapor (Roedder, 1971; Hedenquist and Lowenstern, 1994; Bodnar, 1995). Variations of P, T, and fluid composition, determined by microthermometry on the deposit scale and within one sample, demonstrate the dynamic nature of hydrothermal processes and led to the interpretation of magmatic fluid plumes cooled by external convection of meteoric water (Cathles, 1977; Eastoe, 1978; Fig. 7). Microanalysis of individual fluid inclusions in textural context (Günther et al., 1998; Ulrich et al., 2002; Chang et al., 2018; Jensen et al., 2022) has shown that magmatic fluids prior to ore mineral saturation are much more metal rich than estimates assuming chemical equilibrium with the final ore mineralogy (Kouzmanov and Pokrovski, 2012) and that metals and metalloids are transported by saline liquids as well as vapor-like fluids (Heinrich et al., 1999). However, fluid inclusions can be compositionally modified after entrapment by loss of H2 (preventing deduction of original redox conditions; Mavrogenes and Bodnar, 1994) or by inward diffusion of excess Cu into H2S-rich vapor inclusions (Lerchbaumer and Audétat, 2012; App. VI).

Stable isotopes: O and H isotope analyses of hydrothermal minerals have demonstrated convection of meteoric water around granitic intrusions, whereas the dense network of quartz veins in porphyry deposits is dominated by magmatic fluid (Taylor, 1979; Giggenbach, 1992). Spatially resolved analyses of δ18O of hydrothermal vein quartz indicate that meteoric-magmatic fluid mixing is minor in porphyry Cu-Au deposits, including in the phyllic alteration stage (Hedenquist and Lowenstern, 1994; Watanabe and Hedenquist, 2001; Wilson et al., 2007; Fekete et al., 2016). Isotope ratios of heavier elements like δ65Cu can be interpreted in a similar genetic framework (Gregory and Mathur, 2017).

Laboratory experiments: Measuring fluid properties at high P and T has delineated the conditions of single-phase and two-phase stability of saline fluids and their density and heat content, mainly based on experiments in the H2O-NaCl model system (Sourirajan and Kennedy, 1962; Bodnar, 1995; Driesner and Heinrich, 2007). Experiments also demonstrated considerable metal solubility in liquid as well as vapor fluids (Bischoff and Rosenbauer, 1987). Experiments are key to interpreting fluid inclusions (Steele-Maclnnis, 2018) and larger-scale hydrothermal fluid evolution (Fig. 7). They have quantified how T decrease causes chemical disequilibrium between magmatic rock and fluid derived from the same magma, as the main driving force for mineral precipitation and hydrothermal alteration (Crerar and Barnes, 1976; Lagache and Weisbrod, 1977; Burnham and Ohmoto, 1980; Hemley et al., 1992; Fontboté et al., 2017). Decreasing P at high T makes fluid properties more vapor-like, which weakens the interaction of metal species with H2O molecules and reduces metal solubility (Seward et al., 2014). The solubility of silica as neutral SiO2(aq) also decreases with P and T, except for a region where quartz solubility is retrograde—i.e., increases with fluid cooling through the T interval from ~500° to ~320°C and below ~100 MPa (Fournier, 1983; Akinfiev and Diamond, 2009; Monecke et al., 2018). Sulfur in magmatic-hydrothermal ore fluids acts as additional ligand besides Cl for complexation of Cu (Rempel et al., 2012; Tattitch and Blundy, 2017) and particularly of Au in low-salinity magmatic fluids of near-critical or vapor-like density (Stefánsson and Seward, 2004; Pokrovski et al., 2008). It also contributes S–II for the precipitation of Cu-Fe sulfides, depending on the SO2/H2S ratio in magmatic fluid (Burnham and Ohmoto, 1980; Giggenbach, 1992; Kouzmanov and Pokrovski, 2012). SO2 in fluids expelled by oxidized magmas at depth disproportionates to sulfate and sulfide according to 4S+IVO2+4H2O=H2SII+3S+VIO42+6H+ (Rye et al., 1992). This promotes hydrothermal anhydrite at the expense of the Ca component of plagioclase during potassic alteration (Henley et al., 2015). Upon further cooling, the acidity produced by S disproportionation is the main driver for feldspar hydrolysis to sheet silicates in porphyry Cu systems (phyllic and argillic alteration; Meyer and Hemley, 1959; Hemley et al., 1992). Despite a large database of experimental data in chemically simple systems, their combination in multicomponent reaction models for ore deposition and wall-rock alteration (Helgeson, 1970; Heinrich, 2005; Reed et al., 2013; Hurtig et al., 2021) is still hampered by the lack of a unified thermodynamic description of dissolved species in fluids from vapor to saline liquids (Driesner, 2013).

Fluid evolution and ore mineral precipitation

Field observations, fluid-inclusion data, and mineral solubility experiments allow a semiquantitative interpretation of hydrothermal fluid evolution in porphyry systems. The following interpretation is tentative but consistent with physical constraints imposed by mass and heat flow and with limited thermodynamic reaction modeling. It explores different fluid evolution paths resulting from different T-P gradients between the magmatic fluid source and the surface. Fluid evolution paths change with time as the fluid-producing magma cools and retracts to greater depth (Williams-Jones and Heinrich, 2007; Monecke et al., 2018; cf. Fig. 7).

Changing fluid properties are explained with cartoons illustrating vertical sections through different ore-depositing vein systems (Fig. 10; see App. VI for experimental basis, fluid phase definitions, and further discussion). The schematic channelways depict the state of end-member magmatic fluids, as they ascend and lose heat to external rocks and convecting meteoric water. Stippled red lines schematically indicate positions of isotherms relative to the surface and to the magmatic fluid source. Variation of fluid density is indicated by darkness of color, from gas-like (white) to liquid (dark), and salinity is indicated by the intensity of blue color. This representation allows depicting changes from single-phase (“supercritical”) fluid to two-phase fluid coexistence based on experimental phase relationships in the NaCl-H2O model system. Phase separation may occur in two different ways, depending on how the fluid ascent path intersects the two-phase boundary (vapor + liquid solvus) of the salt-water system. Most magmatic fluids have low salinity (<10 wt % NaCl equiv) and intersect the two-phase boundary on the vapor side. This leads to condensation—i.e., the formation of liquid droplets in a vapor-like fluid. Pure H2O condenses at surface pressure from salt-free vapor cooling below 100°C (Fig. 10A), and the same process occurs if a hotter and denser magmatic vapor containing some salt condenses at higher P. Salt partitions preferentially to the condensing liquid (high-salinity brine), shown as dark-blue droplets in Figure 10B and C, and thereby reduces the salt content of the remaining vapor, whereas gas species such as H2S and SO2 remain in the vapor. Alternatively, phase separation may occur by formation of bubbles in a liquid, if the two-phase surface is intersected on the liquid side, analogous to boiling of liquid water at low P (white bubbles forming in a dense black liquid; Fig. 10C, D). By contrast, fluid contraction, as the opposite of expansion, simply describes density increase or decrease at constant fluid composition. Fluid pressure (Pfluid) not only depends on depth but also varies from lithostatic in the magma to hydrostatic closer to surface. Constrictions in the schematic vein column indicate points of steep Pfluid gradient, controlled by the interplay between hydraulic fracturing and vein closure by plastic deformation and mineral precipitation (Fournier, 1999; Weis et al., 2012).

Four scenarios of fluid evolution from the magma to the surface are expected to have different effects on Cu and Au transport and precipitation (red and orange labels; Fig. 10), and fluid properties at elevated T can be linked with the microscopic appearance of fluid inclusions at laboratory conditions after cooling and exposure (small photomicrographs).

If magma exsolves fluid close to the surface (Fig. 10A), the resulting volatile phase has very low density at near-magmatic T and may be expelled as a volcanic gas by a hot fumarole. Besides H2O, this vapor efficiently extracts gas-like components from the magma, notably SO2. If the vapor cools before reaching the surface, it may condense and mix with meteoric water. SO2 + H2O produces acid (H2SO4) causing feldspar-destructive alteration to clay minerals (argillic alteration), stabilizes sulfate minerals like alunite (advanced argillic alteration), or leaches all rock components except for residual quartz (Rye et al., 1992; Hedenquist et al., 1994; Hedenquist and Arribas, 2022; Fig. 9A). Owing to its low density, the vapor has limited capacity for extracting metals or salts (although metal ratios in high-T volcanic gases are comparable to those in ore-forming fluids, and some sulfides can precipitate as sublimate crystals in hot fumarole conduits; Africano et al., 2002; Nadeau et al., 2010; Edmonds et al., 2018). Further down in the edifice of some volcanoes, or in the roof of ignimbrite reservoirs, NaCl may combine with other salts (KCl, FeCl2) to an H2O-poor salt melt coexisting with low-density vapor (Fiedrich et al., 2020). Focusing of similar fluids forms banded quartz veinlets in sulfide-poor porphyry Au deposits (Fig. 9B; Vila and Sillitoe, 1991; Muntean and Einaudi, 2000; Koděra et al., 2014; Baker et al., 2016). Such veins at Kisladag (Turkey) contain ubiquitous salt melt + low-density vapor inclusions (small photomicrographs in Fig. 10A), and similar salt melts at Biely Vrch (Slovakia) contain measurable concentrations of Au (Koděra et al., 2014). Low-density vapor has a low partial pressure of H2S, inhibiting the precipitation of sulfide minerals such as CuFeS2 or Cu3AsS4 (enargite). As a result, Au ± magnetite dominate the ore mineralogy of most porphyry Au deposits (Muntean and Einaudi, 2000; Fig. 9B). Any Cu (indeed present in the magma, as indicated by Cu-bearing magmatic sulfides; Rottier et al., 2020) only precipitates at greater depth or may become dispersed by meteoric fluids.

Fluid cooling at higher P is favored by exsolution from magma at greater depth (Fig. 10B). Fluid exsolving at 5 to 10 km is denser and able to extract and transport NaCl and ore metals (Fig. 6B), as indicated in Figure 10B through D by increasingly darker and bluer fluid leaving the magma. Cooling of single-phase vapor can lead to contrasting fluid evolutions depending on their initial H2O/NaCl ratio and their P-T path.

If T remains high at rather low P, the vapor expands and its capability to hold salts in solution diminishes during ascent. This forces separation of a highly saline liquid phase (brine or H2O-poor salt melt; Mernagh and Mavrogenes, 2019), condensing from the cooling and expanding vapor, and may lead to saturation with solid NaCl (Lecumberri-Sanchez et al., 2015; dark-blue droplets ± halite indicated in Fig. 10B). If the volumetrically minor brine is trapped in a fluid inclusion, its high salinity is apparent from multiple large salt crystals after cooling. Inclusions of the coexisting vapor show a large bubble and a small meniscus of liquid, as a measure of the original low bulk density (photomicrographs in Fig. 10B from Bingham Canyon). Expansion of the dominant hot vapor with decreasing P tends to favor coprecipitation of ore metals—of Au owing to decreasing fluid density, and of Cu-Fe sulfides and molybdenite while Pfluid is high enough to enable adequate H2S fugacity for sulfide stabilization (Hurtig et al., 2021). Coprecipitation of Cu + Au by an upward-expanding, vapor-dominated fluid plume was proposed following detailed fluid inclusion study at Bingham Canyon. This included mapping depth variations of fluid inclusion density, which correlates with downward-decreasing bulk Au/Cu ratio within the ore shell (Landtwing et al., 2010). A similar process is indicated at the Grasberg Cu-Au deposit (Mernagh et al., 2020). The fluid expansion paths A and B are favored by magmas with rather low H2O content, which explains why Au-rich porphyry deposits are commonly associated with shallow deposits (Murakami et al., 2010) and less fractionated magmas (Sillitoe, 1997; Seedorff et al., 2005).

Somewhat higher P of magmatic fluid exsolution, or more efficient cooling of the magmatic fluid at greater depth, can switch the evolution path to a distinctly different fluid behavior (Fig. 10C). After condensation of brine at depth similar to the previous scenario, the vapor may separate from denser and more viscous brine and ascend to a cooler region, whereby its density increases toward liquid-like properties. This contraction from vapor-like to liquid-like density occurs at constant composition without crossing any phase boundary (Heinrich, 2005). Single-phase magmatic fluids readily transport Au as well as Cu in high concentrations, and phase separation into brine and vapor at near-magmatic T tends to enrich both metals in the more saline liquid phase by forming stable Cl complexes (Zajacz et al., 2017). At intermediate T, however, a P-T window exists where Cl-complexed Cu prefers the saline liquid, whereas Au partitions to a significant degree into coexisting H2S-rich magmatic vapor (Pokrovski et al., 2008). Experimental data defining this window are ambiguous, but in aqueous liquids of near-critical density at 400° to 500°C, Au bisulfide complexes are stable enough to permit ppm-level gold solubilities (Stefánsson and Seward, 2004). Brine-vapor separation followed by contraction of the ascending vapor can, therefore, lead to segregation of Cu into the brine (available for Cu-Fe sulfide precipitation in deeper parts of zoned porphyry deposits like Bingham Canyon; Landtwing et al., 2010) from Au transported with the contracting vapor into overlying epithermal gold deposits (including Carlin-type; Sillitoe and Bonham, 1990; Muntean et al., 2011; Large et al., 2016). This interpretation modifies earlier thermodynamic modeling in light of new fluid inclusion observations (Lerchbaumer and Audétat, 2012) and explains the separation of Au from Cu in agreement with the low Cu content of most epithermal gold deposits. Brine-vapor separation at depth favors Au transport to epithermal temperatures, because it depletes the vapor phase in FeCl2, helping to maintain high H2S concentrations in aqueous liquids needed for Au complexation (Heinrich, 2005; see App. VI for geologic evidence and further discussion). Conversely, pressure-dependent separation of Au and Cu can explain the vertical zonation of ore metals at Bingham Canyon, with Au-deficient but Cu-rich lobes at depth and high-grade Cu + Au coprecipitating in the shallow apex of this zoned ore shell (Landtwing et al., 2010).

Vapor expansion (Fig. 10B) may change with time to vapor contraction (Fig. 10C), observing a given point in an evolving orebody. Batu Hijau (Indonesia) shows high-grade bornite deposition during the first pulse of porphyry emplacement and veining in parts of the deposit, together with biotite-feldspar alteration (Setyandhaka et al., 2008). Quartz-bornite veins formed at high T with feldspathic alteration, perhaps initially by decompression and expansion of low-salinity single-phase fluid to brine and vapor. Quartz veins related to the second and main mineralizing porphyry intrusion contain Cu-Fe sulfides in later fractures, which Cernuschi et al. (2023) interpreted as a result of sulfide remobilization. Alternatively, lower-T bornite and chalcopyrite may directly precipitate in later fractures overprinting higher-T quartz veins from a single-phase aqueous liquid at 300° to 400°C. Textures and inclusion analyses suggest that this fluid formed by rehomogenization of magmatic brine (providing most Cu ± Au) and vapor (rich in H2S) that had previously separated slightly deeper along the upflow path through the deposit (Schirra et al., 2022). This time-space evolution is expected from overall cooling of the magmatic system, leading to retraction of isotherms, of the brittle-ductile transition, and of the magmatic/meteoric fluid front to greater depth with time. As a result, advected ore metals are accumulated in a downward-growing ore shell, as predicted by Weis et al. (2012; Fig. 7). A similar evolution was proposed by Hedenquist et al. (1998) for the Far Southeast porphyry Cu-Au deposit and coeval high-sulfidation Au-Cu-As veins at Lepanto (Philippines), starting with early quartz veins at depth and vuggy quartz alteration in the shallow vein system driven dominantly by expanding vapor. This was followed by chalcopyrite addition with sericite alteration in the deeper porphyry and Au + pyrite + enargite addition to shallow epithermal ore by cooling aqueous liquid.

Fluid production at even higher P (Fig. 10D) favors cooling of the fluid without intersection of the two-phase brine + vapor region, as exemplified by the Butte porphyry Cu-Mo deposit (Rusk et al., 2008; Reed and Dilles, 2020). Copper mineralization is associated with biotite-K-feldspar-muscovite alteration grading upward to muscovite-chlorite halos containing most of the chalcopyrite of the deposit. Fluid inclusions in early dark micaceous veins as well as barren quartz and quartz-molybdenite veins (Fig. 9E) have low salinity, relatively high CO2 content, and intermediate density, indicating that porphyry-style ore mainly formed from a single-phase fluid between 650° and 475°C (Rusk et al., 2008). Pressure estimates from fluid inclusions indicate that mineralization depth decreased from 9 to 2.5 km, consistent with geologic reconstructions indicating surface erosion during protracted ore formation, with crosscutting epithermal sulfide veins extending laterally away from the porphyry Cu-Mo orebodies (Reed and Dilles, 2020). No information exists about the original Au endowment, but high fluid density, high H2S content, and feldspar-destructive alteration would counteract precipitation of any contained Au and favor its removal into the now-eroded near-surface domain.

Alternative arguments and open questions

Variations in fluid properties during the cooling of large igneous magma system are the prime control on hydrothermal alteration and mineral precipitation at the deposit to vein scale. The following paragraphs point out a few limitations and alternative arguments to this first-order interpretation (see App. VI for additional discussion).

Cu/Au ratio: The bulk Cu/Au ratio of porphyry deposits tends to increase with depth and P of ore deposition (Murakami et al., 2010) but is first of all limited by the availability of metals in the input fluid and its source magma, preceded by melt formation and lower crustal differentiation. This is shown by Au-poor deposits that formed at rather low P (e.g., El Teniente; Klemm et al., 2007) and possibly deeper porphyry-style Au deposits without reported salt melts that formed from distinct magmas or by different precipitation controls (e.g., La Colosa, Colombia; Sillitoe, 2017; Naranjo et al., 2018)

Au in bornite solid solution: Precipitation of Au as a trace component dissolved in bornite was proposed as a dominantly mineralogical control on Au-rich porphyry deposits (Kesler et al., 2002). This interpretation is linked to the debate about the temperature of original Cu-Fe sulfide precipitation. For a porphyry Cu-Au deposit with high bulk Au/Cu ratio, such as 1 ppm Au per 1 wt % Cu at Grasberg, at least 560°C is required to initially incorporate all Au into bornite solid solution, or in excess of 660°C if it was originally dissolved in chalcopyrite. This may be possible in the T range of ore deposition proposed by Cernuschi et al. (2023), implying that Au was later exsolved to separate grains now making up the bulk of the gold content of the ores. Lower temperature of Cu-Fe sulfide deposition, as proposed by Landtwing et al. (2010), Schirra et al. (2022), and other studies, necessitates that Au is coprecipitated as native Au grains together with Cu-Fe sulfide, as the two metals occur today.

Vapor or brine dominance: The relative contribution of two fluid phases to porphyry mineralization is a matter of definition and varies in space and time. In most deposits, the input ore fluid is a single phase that commonly intersects the salt-water two-phase curve on the vapor side of the solvus, so it is “vapor-like,” having a density lower than the critical density of a fluid of this composition. Phase and element proportions after phase separation of the single-phase ore fluid to brine + vapor can be calculated based on equilibrium experiments in the NaCl-H2O-(-CO2) model system (invariably showing a vapor/brine mass ratio >1) and fluid inclusion analyses of ore metals (Landtwing et al., 2010; Chang et al., 2018; Schirra et al., 2022). Such closed-system mass balances variably conclude that Cu dominantly resided in the brine (Yulong, Tibet; Batu Hijau) or existed in comparable fractions in brine and vapor (Bingham Canyon; after correction for Cu overestimation due to postentrapment Cu gain). Apart from severe analytical uncertainties in such calculations, the distribution of ore-forming components between two fluid phases only becomes important if vapor physically separates from brine on the scale of the fluid plume (Chang et al., 2018) but is immaterial if the two fluid phases flow together through the vein system and later rehomogenize (Schirra et al., 2022).

Brine lenses: Accumulation or “ponding” of Cu-rich brine is unlikely to contribute decisively to ore formation, even though brine lenses occupying pore space in and around shallow magma conduits are geophysically observed below active volcanoes (Afanasyev et al., 2018). Even if the Cu concentration in the fluid were 1 wt % Cu, as in the most Cu-rich brine inclusions (Kouzmanov and Pokrovski, 2012; Audétat, 2019), the formation of ore with 1% Cu grade would require that the rock at time of mineralization consisted of ~60 vol % fluid and ~40 vol % host rock (assuming that the brine had a density of 1,100 kg/m3 and the original host rock a density of 2,700 kg/m3). Even with less extreme assumptions, accumulation of metal-rich brine is not considered to help economic ore precipitation, as copious sulfur originally present in the lower-salinity ore fluid already escaped with the dominant vapor phase.

Heat budget: The thermal consequences of focused fluid flow, including contraction of magmatic vapor and its mixing with meteoric water, are not yet fully understood. Multiphysics models explain key features of porphyry ore shells but do not include the initiation of the fluid plume and the emplacement of a hot porphyry into an initially cold near-surface environment. The heat content of the magma is small compared with the heat advected by fluid during mineralization but may not be negligible (Cathles, 1977), especially if the three-dimensional contact between a thin porphyry finger and cold water-saturated host rock is considered. Initially steep T gradients together with high Cu concentrations in the first fluid batches (Fig. 6) could explain the observation that high-grade bornite mineralization in early quartz veins best develops with the first mineralizing porphyry (e.g., Batu Hijau; Setyandhaka et al., 2008; Proffett, 2009), prior to establishment of a stable fluid plume accumulating the prominent ore shell (Schirra et al., 2022). In addition to changes in P, T, and fluid density, high-grade Cu deposition is also influenced by wall-rock chemistry—e.g., skarns promoting acid neutralization or Fe-rich host rocks stabilizing Cu-Fe sulfides (Sillitoe, 2010)—but even these alteration reactions are driven by temperature decrease.

Porphyry Cu deposits forming at several kilometers depth in orogenic settings are subject to later exhumation, explaining the characteristic time difference between formation and exposure at the surface. This delay is on the order of tens to hundreds of million years in global average; much older deposits require special conditions for their preservation near the present-day surface. Statistical treatment of the age distribution of exposed deposits allows estimation of the number of undiscovered deposits still existing under >1 km of barren overburden but shallow enough to be accessible to present mining technology (~3 km). Such models indicate that potentially mineable Cu resources are several orders of magnitude larger than estimates based on exposed deposits (Kesler and Wilkinson, 2008).

Many porphyry Cu deposits have low Cu grade in hydrothermally altered rock and become economic only thanks to supergene enrichment during exposure to the hydrosphere. Supergene enrichment involves dissolution of primary Cu-Fe sulfides by surface water, by oxidizing sulfide to sulfuric acid and Cu+I to Cu+II, which are both highly soluble at atmospheric oxygen level. Cu is reprecipitated when oxidized acid groundwater contacts hydrothermal sulfides lower in the weathering profile, reducing Cu+II to secondary sulfides like Cu2S (chalcocite). Alternatively, Cu+II silicates, carbonates, or sulfates can accumulate when the acid groundwater contacts pH-neutralizing feldspar or carbonate rocks. Inorganic pyrite oxidation is kinetically slow, and even over geologic time periods only occurs thanks to bacterial catalysis (Sillitoe, 2005, with earlier references). Space and lack of expertise preclude a deeper review here, even though the process is an interesting link between economic geology and tectonics, mountain building, and climate evolution (e.g., Dahlstrom et al., 2022). Supergene enrichment also has environmental advantages during processing, as slow natural dispersion of toxic elements over geologic timescales avoids environmental challenges associated with rapid sulfide mining and long-term storage of reactive waste (Arndt et al., 2017).

Understanding geologic process is one of many tools for meeting the needs of future metal supply, and rigorous physics and chemistry are essential for interpreting seemingly obvious geologic observations. Academic progress is slow, but common knowledge today was forefront research in the past—notably the understanding of hydrothermal alteration some 50 years ago. We do not know which of the more recent academic research is useful in the future. Whichever, I believe that results of technically complex research should be explained as intuitive concepts that can be put to the test by practitioners engaged in the discovery of new metal resources. Here I list some conclusions that may be significant for exploration and some questions that I consider particularly relevant for future research.

Does global-scale heterogeneity of the lithosphere predispose Cu-, Mo-, or Au-rich porphyry provinces? Metal endowment, notably Mo/Cu and Au/Cu, is partly determined by plate tectonics proceeding over Earth-scale periods of mantle preparation since the Precambrian but may also be affected by changes in plate configuration lasting a few million years or less. The initial Au content of primitive basalts is plausibly related to preenrichment of the mantle, in absolute Au concentration or relative to Cu, prior to selective mobilization by a later event of mantle remelting. Likewise, two global Mo provinces seem to be related to an unknown process of selective metal enrichment during the early Precambrian. The initial Cu content of mantle magmas is generally posited to be constant, but mantle magmas selectively enriched in Cu might be masked by upper crustal processes of Cu loss prior to sampling and rock analysis. Source enrichment of Cu in certain mantle domains should not yet be excluded as a factor controlling later generation of unique Cu provinces such as the central Andes, even if evidence and possible mechanisms are presently elusive.

Recognizing fertile magma systems within known provinces and discovering fertile areas in older terrains reworked by later metamorphism are essential for future discovery. Zircon is particularly useful because it survives later metamorphism as well as weathering and, being a chronometer as well as chemical indicator, allows identification of brief periods of magma fertility in longer periods of unproductive magmatism. Since deep fractionation to hydrous magmas is a key to magma fertility, a chemical indicator measuring lithostatic P (independently from H2O fugacity) during zircon crystallization would be high on my wish list for research and would soon be of practical value in mineral exploration.

What is the geometry of a successful magma plumbing system able to generate a hydrothermal fluid plume fast enough to develop the thermal anomaly needed for making a giant ore deposit? This question may be the most important point of debate in the present state of porphyry Cu research. Whether pipe-like highways transfer exceptionally hydrous magma directly from the lower to the upper crust or whether moderately rapid magma flow fills exceptionally large but nonerupting magma chambers in the upper crust is not just an academic question. Recognizing one or the other case in exploration terrains must be a first-order exploration criterion. Should explorers look for evidence of large upper crustal plutons and develop methods for mapping their internal structure and remotely identifying former points of fluid discharge? Or should they develop methods for identifying steep magma channelways of great vertical extent?

The nature of mineralization pulses may commonly be misinterpreted. Heat balance calculation and hydrothermal alteration sequence show that fluid flux is probably more continuous on the scale of an orebody than sharply intersecting veins and timelines of intermittent porphyry emplacement may suggest. Not the volumetrically small porphyry intrusions but the sustained fluid flux is key to accumulating metals in a finite rock volume, as opposed to metal dispersion over larger distances either along the plumbing system or in multiple pulses extending over long time periods. High ore grade results from the balance between hot magmatic fluid flux and heat removal by an effective cooling engine. Repeated addition of porphyry intrusions is not beneficial as such and is more likely to dilute than to enhance ore grade.

Different process durations at different scales are a key to deposit size and ore grade, and cautious interpretation of high-precision geochronological data based on critical sampling is a promising discriminator for recognizing potentially large and rich deposits. However, a larger spread of geochronological ages alone does not necessarily indicate a better ore system. Rather, physical modeling and geochronology indicate that the best porphyry deposits form as fast as they can but take as long as needed for producing the required ore fluid from a mass of hydrous magma.

The temperature anomaly during ore formation is the most diagnostic rock signal extending beyond the small footprint of an economic orebody. It can be detected by alteration and mineral-compositional mapping or by fluid inclusion observation even in poorly exposed river catchments. Fluid inclusions do not require academic sophistication to be used in exploration practice. During early drilling, simple inspection of bubble/liquid/halite ratios can inform about formation depth and potential metal zonation.

This paper is a late derivation from an SEG Distinguished Lecture, and I would like to thank Larry Meinert, the editor of Economic Geology, for inviting this review and for his patience. The article represents a personal assessment of three decades of research on magmatic-hydrothermal and other ore systems. Much of this research was completed by my colleagues and our joint students. These include scientists from neighboring fields like physics and chemistry, whom ETH Zürich and the Swiss National Science Foundation supported as a diverse team over a long period. I name only three founding members of our group, including Thomas Driesner, Detlef Günther, and Albrecht von Quadt, who had a decisive influence on later developments, and a few recent postdocs and graduate students, including Yannick Buret, Benoit Lamy-Chappuis, Simon Large, and Michael Schirra, whose results particularly contributed to the thoughts expressed in this paper. Exploration and mining geologists supported our students with their hospitality, with practical questions, and with access to invaluable data and sample material. Among these industry colleagues, the late Vic Wall was a particularly stimulating supporter of our group. I have learned about porphyry systems from many academic colleagues among whom, in hindsight, Marco Einaudi was perhaps most influential. I am most grateful to these people and many unnamed students, collaborators, and academic colleagues. The paper has benefited from formal and informal reviews by Andreas Audétat, Cyril Chelle-Michou, David Cooke, John Dilles, Thomas Driesner, Robert Loucks, Chetan Nathwani, John Proffett, Eric Seedorff, Richard Sillitoe, Thierry Solms, Derek Vance, and Zoltán Zajacz and from critical input by many others. This article summarizes research results obtained thanks to long-term support by the Swiss National Science Foundation through grant 200020_166151.

Christoph A. Heinrich studied metamorphic petrology in Switzerland, before moving to Australia to follow his interests in economic geology with Comalco, the Commonwealth Scientific and Industrial Research Organisation (CSIRO), and Geoscience Australia. Partly supported by industry, he investigated the metasediment-hosted copper system of Mount Isa, with a stint into experimental hydrothermal geochemistry at the Department of Scientific and Industrial Research (DSIR) in New Zealand. In 1994 he was appointed professor of mineral resources at ETH Zurich. Here he built a teaching and research group investigating processes of ore formation, combining field studies with new analytical and modeling methods. In 2017 he was awarded the Penrose Gold Medal of the Society of Economic Geologists. As emeritus professor, Chris now works as an independent researcher and geologic consultant.

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.