Orogenic gold deposits contribute the largest proportion of the world’s gold reserves, and the source of their ore-forming components has been recognized as the metamorphic devolatilization of metapelites or metabasites across the greenschist- to amphibolite-facies transition. However, hypozonal orogenic gold deposits represent an enigma in this context. Some of these apparently formed in higher-grade metamorphic rocks when temperatures were beyond the wet solidus of quartz-feldspar–bearing rocks; it is therefore puzzling how these fluids were generated in the source and migrated through the crust without causing partial melting. Here, we show that devolatilization of hydrated komatiites, a volumetrically significant lithological unit in Precambrian greenstone belts, is a viable model that can plausibly lead to gold mineralization at amphibolite-facies conditions. Our thermodynamic simulations indicate that subsolidus metamorphic devolatilization of komatiites at ~700 °C (upper amphibolite facies) can unlock significant amounts of gold via dehydration of talc and chlorite. This genetic model is supported by the geochemical characteristics of, and estimated pressure-temperature (P-T) formation conditions of, hypozonal gold deposits and the intimate spatiotemporal association between hypozonal deposits and komatiites in greenstone belts. This work expands the P-T range of the metamorphic devolatilization model and enhances its robustness in explaining gold mineralization in metamorphic terranes.

Orogenic gold deposits are widely studied because they contribute over a quarter of the world’s gold supply (Goldfarb et al., 2005); however, the source(s) of their ore-forming components (fluid, sulfur, gold, and other metals) has(have) long been debated (Goldfarb and Groves, 2015; Groves et al., 2020; Kolb et al., 2015; Phillips and Powell, 2010; Selvaraja et al., 2017; Tomkins, 2010; Wang et al., 2022; Zhao et al., 2019). The widely accepted metamorphic devolatilization model (Phillips and Powell, 2010) emphasizes that gold-bearing fluids are produced by metamorphic dehydration of hydrous crustal rocks, particularly at the greenschist to amphibolite transition, largely through breakdown of chlorite (~12% H2O) to minerals like biotite (~4% H2O), hornblende (~2% H2O), and garnet (anhydrous) (Goldfarb et al., 2005; Phillips and Powell, 2010; Pitcairn et al., 2006; Tomkins, 2010; Zhong et al., 2015). At temperatures higher than the greenschist-amphibolite transition, there is minimal opportunity for fluid liberation from the metamorphosed mafic and sedimentary rocks. As a result, gold and sulfur are thought to be inaccessible in the sources under these conditions (Tomkins, 2013). Since fluids generated at the greenschist-amphibolite transition tend to migrate upward into rocks of lower metamorphic grades, this model satisfactorily explains the formation of orogenic gold in lower-amphibolite- to greenschist-facies terranes.

Deposits are also found in higher-grade metamorphic rocks, known as hypozonal gold deposits. These deposits are mostly hosted in amphibolite-facies rocks in greenstone belts and are characterized by high-temperature alteration assemblages such as garnet, diopside, hornblende, and K-feldspar and ore-forming temperatures as high as ~500–700 °C (Kolb et al., 2015). Although several cases are thought to have formed before peak metamorphism and were later overprinted by high-temperature metamorphism (e.g., Big Bell—Phillips and Nooy, 1988; Challenger—Tomkins and Mavrogenes, 2002; Griffin’s Find—Tomkins and Grundy, 2009; Hemlo—Tomkins et al., 2004; Glenburgh—Roche et al., 2017), most of the hypozonal deposits have been proven to have formed simultaneously with, or slightly later than, the peak metamorphism in the ore-hosting terranes (Kolb et al., 2015; Groves et al., 2020). Cases studies on some of these deposits (e.g., Chalice and Three Mile Hill in the Yilgarn Craton) also suggest a metamorphic origin for their ore-forming components (Selvaraja et al., 2017). Some of these deposits may be plausibly explained by the thrust-over model, where fluids generated from the underlying still-hydrous greenschist-facies rocks migrate upward to hotter higher-grade rocks when the latter thrust over the former (e.g., Renco—Blenkinsop and Frei, 1996), but this cannot explain why in some cases voluminous fluids can pass through the overlying rock piles at temperatures higher than the wet solidus of quartz-feldspar–bearing rocks (~650 °C; Tomkins and Grundy, 2009). In addition, in Archean greenstone belts globally, hypozonal gold deposits are found in transpressional greenstone belts, where there is limited opportunity for thrusting of amphibolite-facies rocks over large volumes of greenschist-facies material. Therefore, the presence of hypozonal deposits creates a conundrum with regard to fluid source. It is noted that the hypozonal deposits discussed in this article are restricted to those formed generally coeval with host-rock metamorphism, and they do not include Mesozoic gold deposits in Jiaodong or elsewhere in the North China block, which are hosted in high-grade basement rocks but formed ~2 b.y. postdating host-rock metamorphism (Groves et al., 2020; Kolb et al., 2015).

In addition to metasedimentary and metabasaltic rocks, komatiites are important components in Precambrian greenstone belts and have been regarded as the potential gold source for some Archean gold deposits (Keays, 1984). Komatiites are ultramafic lava flows that originate from the mantle with high MgO contents and similar components as peridotite (mainly olivine and pyroxene). The lower viscosities of komatiite lavas determine that they are easily erupted onto the seafloor and readily undergo alteration by seawater (Tamblyn et al., 2023). The hydrated komatiites undergo metamorphic devolatilization through a series of dehydration reactions, some of which take place at temperatures beyond the greenschist- to amphibolite-facies transition (e.g., chlorite breakdown at ~800 °C; Hartnady et al., 2022). They also have a considerably higher wet solidus temperature than quartz-feldspar–bearing rocks (~1000 °C; Hartnady et al., 2022). Here, we simulated devolatilization of gold- and sulfur-bearing metabasite and komatiite to evaluate the mobility of gold during metamorphism of greenstone belts (see Methods in Supplemental Material File S11). The bulk-rock compositions of metabasite and komatiite used for simulation were based on a compilation of rock geochemistry in greenstone belts worldwide, and native gold was assumed to be the only gold-bearing mineral in these rocks, according to geological observations showing that gold in these rocks mainly exists as native gold or Au-Ag alloys (Zelenski et al., 2017; Dare et al., 2010; see Methods in Supplemental Material File S1).

Gold Release during Devolatilization of Metabasites

The simulated metamorphic mineral assemblages are consistent with petrological observations (Manning et al., 1993), which are characterized by actinolite-epidote-chlorite-plagioclase assemblages at greenschist facies and hornblende-plagioclase-chlorite at amphibolite facies. The simulated amounts of water liberation (Fig. 1) are also in good agreement with previous models (Elmer et al., 2006; Powell et al., 1991). Auriferous metamorphic fluids are mainly liberated by decomposition of chlorite and epidote during the transition from greenschist to amphibolite facies (450–550 °C), where up to ~2 wt% of free water relative to rock mass can be released (Fig. 1A). Coupled with fluid liberation at the greenschist-amphibolite transition, ~1–1.5 ppb of Au (relative to rock mass) can be scavenged from the source rock, mainly as Au(HS)2 in the fluid (Figs. 1C and 1D). This indicates that the majority of Au (50%–70%) in basaltic rocks will be stripped at the greenschist-amphibolite transition (assuming Archean basaltic rocks contain ~2 ppb Au; Supplemental Material File S2). This simulation result is consistent with the systematic gold loss in variably metamorphosed metabasites observed at La Grande and the Central Lapland greenstone belts (Patten et al., 2020). At the greenschist- to amphibolite-facies transition, the gold-releasing process is more effective at hotter geothermal gradients (Fig. 1C), supporting the suggestion that hot orogens are favorable for orogenic gold mineralization (Phillips and Powell, 1993; Tomkins, 2010). After entering the amphibolite facies, very limited amounts of metamorphic fluids and gold can be released, and the metamorphic devolatilization process ceases at temperatures higher than the wet solidus of basaltic rocks (Figs. 1A and 1C). Such a pattern of fluid and gold liberation is similar to devolatilization of metapelites (Zhong et al., 2015), and thus these two rock types contemporaneously liberate gold-bearing fluid, consistent with the classical orogenic gold model.

Gold Release during Devolatilization of Komatiites

Metamorphosed komatiites are characterized by chlorite-antigorite-olivine assemblages at greenschist facies and chlorite-talc-anthophyllite-olivine assemblages at amphibolite facies, consistent with petrological observations of komatiites in greenstone belts (Gole et al., 1987). Devolatilization of hydrated komatiites produces two steps of fluid and gold liberation (Fig. 2). First, when crossing the greenschist-amphibolite transition at ~550 °C, breakdown of antigorite releases ~4 wt% H2O and ~1–1.4 ppb Au (relative to rock mass), mainly as Au(HS)2, in the fluid (Figs. 2A, 2C, and 2D). After entering the amphibolite facies, a sharp rise in gold liberation occurs at ~700 °C in response to dehydration of talc and chlorite, which liberates ~0.5 wt% of H2O and ~2–5.5 ppb of Au (Figs. 2A, 2C, and 2D), mainly as Au(Cl)2. Although the second step of devolatilization releases less H2O than the first, more gold can be scavenged due to the higher solubility of gold at higher temperatures. Within the pressure-temperature (P-T) regime of the greenschist-amphibolite transition, the solubility of gold is ~0.02–0.06 ppm in metamorphic fluids, and this builds to ~0.4–1 ppm as temperatures increase to ~700 °C (Fig. 2D).

At ~700 °C, devolatilization can cumulatively unlock ~3–6 ppb of Au relative to the rock mass (Fig. 2C). Considering that ~1–1.5 ppb of Au can be cumulatively released from metabasites (Fig. 1C) and ~2 ppb can be released from metapelites (Zhong et al., 2015), komatiites can potentially provide more gold than metabasites and metapelites given the same rock volumes, because of the higher solubility of gold in amphibolite-facies fluids. Based on geochemical data from the Yilgarn Craton (see Methods in Supplemental Material File S1), most komatiites have been intensively hydrated and have ~6–14 wt% mineralogically bound water (Fig. S1; Supplemental Material File S3). In addition, these komatiites have significantly higher Au contents (~3–7.5 ppb median) than basaltic rocks (~2 ppb) and thus can provide sufficient gold during high-grade metamorphism (Fig. S2; Supplemental Material File S2), provided that fluids can transport and then precipitate gold at a structurally focused location.

Our contention that hypozonal gold deposits can be sourced by metamorphic devolatilization of komatiites at upper-amphibolite-facies conditions (~700 °C) is supported by several geological observations. Compared to mesozonal gold deposits, the alteration and mineralization of hypozonal gold deposits are characterized by less silica enrichment and greater abundances of Ni, Co, S, and Se, reflected by the common presence of skarn-like alteration assemblages and massive or semimassive sulfide mineralization (Kolb et al., 2015). These signatures may reflect the geochemical fingerprints of komatiites, which are silica-undersaturated ultramafic rocks enriched in compatible elements such as Co and Ni. The stronger enrichment in S and Se can be explained by the higher sulfur contents (~100.5 molal; Fig. 2B) in komatiite-sourced fluids at ~700 °C than contents from basalt- or pelite-sourced fluids at the greenschist-amphibolite transition (~10−1 molal; figs. 1b and 2b in Zhong et al., 2015). Given that Se does not fractionate from S in most geological processes (Alirezaei and Cameron, 2001), its enrichment is expected in these S-rich fluids.

Based on a compilation of the most well-studied hypozonal gold deposits worldwide, Kolb et al. (2015) found that their ore-forming P-T conditions define a linear trend with the highest end point at ~700 °C and 7 kbar. Kolb et al. (2015) suggested that this end point represents the P-T limit of the fluid sources for all hypozonal gold deposits, although the nature of this inferred source was unknown. We suggest that the komatiite devolatilization model provides the solution, characterized by a very narrow and rich window of gold liberation at ~700 °C (Figs. 2A and 2C). At higher temperatures, very limited amounts of gold can be further released from komatiites (Fig. 2C), consistent with the 700 °C upper limit of Kolb et al. (2015). In addition, komatiites would have undergone up to ~7 ppb of gold loss by the time they reached 700 °C (Fig. 2C), and given that they initially have ~3–7.5 ppb gold (Fig. S2), they would typically be gold depleted beyond ~700 °C.

The higher wet solidus temperature of komatiites, and to a lesser extent mafic rocks, provides a solution to the problem of transporting fluids at high temperatures. Since serpentinites accommodate strain more readily than other rocks (Escartín et al., 1997), they tend to deform and create the pathways for fluid migration, allowing fluids to avoid interacting with quartz-feldspar–bearing rocks where they would be consumed by partial melting. Similarly, shear zones developed at the interface between komatiites and mafic rocks are also permissible fluid transmission pathways, particularly as fluids cool below 650 °C. After cooling to subsolidus temperatures of quartz-bearing basaltic or granitic rocks, the fluids would interact with these lithologies, become saturated in silica, and eventually generate mineralization with quartz as a hydrothermal mineral.

The genetic link between hypozonal gold and komatiites is also manifested by their close spatial and temporal associations. In the Yilgarn Craton in western Australia, where hypozonal gold has been extensively studied, hypozonal deposits are found only in high-grade metamorphic regions containing komatiites (e.g., Youanmi and Kalgoorlie terranes; Fig. 3). In contrast, hypozonal gold deposits are absent in high-grade regions lacking komatiites, such as those in the Southwest terrane (Fig. 3). In terms of temporal correlation, hypozonal gold deposits are broadly contemporaneous with komatiites in greenstone belts worldwide (Fig. 4). This correlation between hypozonal mineralization and komatiites explains why the majority of hypozonal deposits are Archean.

Breakdown of chlorite and talc in hydrated komatiites at ~700 °C can release substantial amounts of auriferous fluids without causing partial melting. Combined with the availability of gold and mineralogically bound water in hydrated komatiites, metamorphic devolatilization of komatiites can well account for hypozonal gold mineralization in amphibolite-facies rocks, as well as the close temporal and spatial association between hypozonal gold and komatiites in greenstone belts.

1Supplemental Material. File S1: Detailed description of simulation methods. File S2: Gold contents of volcanic rocks in West Australian Craton. File S3: Bulk rock water contents of komatiites in global greenstone belts. Data for Files S2 and S3 are compiled from GEOROC database. Please visit https://doi.org/10.1130/GEOL.S.24492283 to access the supplemental material; contact [email protected] with any questions.

We thank science editor Marc Norman, Neil Phillips, and two anonymous reviewers for their constructive comments. This work was financially supported by the National Natural Science Foundation of China (42222303, 41872078, 41930427, and 42203064) and the China State Key Research Plan (grant no. 2021YFC2901703).