An intensive study of the geochemical characteristics (including the volatile elements Cl and S) of apatite associated with porphyry deposits was undertaken to address the debate about the crust- or mantle-derivation of their copper and gold and to better understand the controls on the transport of metals in magmatic fluids in post-subduction settings. New geochemical data on apatite reveal parameters to discriminate mineralized porphyry systems across Iran and western China (Tibet and Yunnan), from coeval barren localities across this post-subduction metallogenic belt. Apatites in fertile porphyries have higher Cl and S concentrations (reflecting water-rich crystallization conditions) than those from coeval barren ones. Our new isotopic data also indicate these volatiles are likely derived from pre-enriched sub-continental lithospheric mantle, metasomatized by previous oceanic subduction. This study demonstrates that refertilization of suprasubduction lithospheric mantle during previous collision events is a prerequisite for forming post-subduction fertile porphyries, providing an evidence-based alternative to current ore-enrichment models.

The Earth's mantle and crust contain little copper and gold, and thus the concentration of these elements to form ore deposits requires an explanation of their origin (crust or mantle) and the processes required (Sillitoe 2010). Volatiles such as chlorine, sulfur, and water are of paramount importance for the transport and deposition of copper and gold (Griffin et al. 2013; Tassara et al. 2017) and carry the keys to examining this problem. The Tethyan metallogenic belt, extending from Europe through Iran to Tibet, is one of the world's major metal-producing belts. It contains many post-collisional porphyry deposits and is an ideal natural laboratory to examine the origins of important volatile components and the related mechanisms for post-subduction porphyry mineralization.

Volatiles are difficult to study in whole-rock samples, owing to alteration and weathering. However, apatite concentrates volatiles (e.g., chlorine and sulfur) that are ubiquitous in the lithospheric mantle (O'Reilly and Griffin 2000) and crust (Bruand et al. 2019) and especially apatite crystallizes early in porphyry systems (Mao et al. 2016). Apatite, therefore, can record a more complete history of the evolution of volatiles in these magmas (Chelle-Michou et al. 2017).

In this study, we use in situ major- and trace-element geo-chemistry and Sr isotopes of magmatic apatite in porphyries from 12 porphyry Cu ± Au ± Mo deposits and 7 barren suites within the Tethyan metallogenic belt. The sample suite includes the largest porphyry deposits in each mineralized domain. These data reveal (1) anomalous enrichment of chlorine and sulfur in ore-forming magmas, and (2) a mantle isotopic signature in fertile suites.

Fifty samples from 12 fertile porphyry deposits and 7 barren locations in four major post-subduction arcs (Fig. 1) have been analyzed, including the two largest porphyry systems in Iran, the Sarcheshmeh and Sungun deposits in the Kerman and Arasbaran porphyry copper belt; the deposits in the Gangdese belt, southern Tibet; and the porphyry systems in the Sanjiang orogen in southeast Tibet (Online Material1 Fig. OM1). All samples represent post-collisional magmatism following cessation of active subduction. The description of the deposits and the barren plutons is given in the Online Material1. The apatite samples were collected from fresh magmatic apatite, without zoning and extensive resorption texture from BSE images (Online Material1). To yield robust apatite data, especially for the S, Cl, and isotopic analysis, we carried out 244 major-element and 562 trace-element analysis on seven different apatite standards and synthetic materials, including Durango (Mexico), Mad-blue, Mad-green (Madagascar), Moy (Myanmar), Mud Tank (Australia), and Kovdor and Sly (Russia). Rb-Sr isotopes in apatites with different Sr concentrations were measured by the solution method and then used as monitoring standards. The description of standards and analytical methods is given in the Online Material1.

Chlorine, sulfur, and rare-earth elements

The chlorine contents of apatite from the ore-forming porphyries are generally higher than those of apatite from the barren plutons (fertile suites: 0.08–0.34 wt%; median 0.13 wt%, average 0.17 wt%; barren suites: 0–0.34 wt%; median 0.03 wt%, average 0.04 wt%). For example, apatite in the Sarcheshmeh and Sungun porphyry Cu deposits contains 0.89 to 1.36 wt% and 0.42 to 1.64 wt% Cl, respectively; comparative values are 0.12 to 0.58 wt% Cl in the Qulong porphyry Cu-Mo deposit, ~0.06 to 0.23 wt% Cl in the Jiama porphyry-skarn Cu-Mo-Au deposit, 0.02 to 0.12 wt% Cl in the Beiya porphyry-skarn Cu-Mo-Au deposit, and 0.12 to 0.41 wt% Cl in the Machangqing porphyry deposit (Fig. 2a; Table 1). In contrast, 95% apatites from barren suites contain <0.1 wt% Cl. Sulfur enters the apatite structure by complex substitutions including Na+ + S6+ = Ca2+ + P5+ and S6+ + Si4+ = 2P5+(Sha and Chappell 1999). The SO3 contents in apatite from the fertile suites (<0.96 wt%; median 0.15 wt%, average 0.19 wt%) are higher than those in apatite from the barren suites (<0.75 wt%; median 0.07 wt%, average 0.10 wt%; Fig. 2a). REEs and Y are incorporated into the apatite structure via coupled substitutions (Hughes and Rakovan 2015): Na+ + (Y+REE)3+ = 2 Ca2+ and Si4++ (Y+REE)3+ = P5+ + Ca2+. Apatites in the mineralized samples have higher contents of REEs+Y (28–23 000 ppm; median 520 ppm, average: 6650 ppm) than those in the barren rocks (68–55 400 ppm; median 3230 ppm, average: 4110 ppm; Figs. 2b). The barren and fertile suites can also be distinguished on a series of trace-element plots involving (Ce/Pb)N, (La/Yb)N, and (V/Y) (Fig. 2b). In detail, apatites from most of the fertile-suite samples have higher (La/Yb)N (fertile suite: 2–206; median 67, average 73; barren suite: 0–155; median 18, average 23) and V/Y (fertile suite: <0.4; median 0.04, average 0.05; barren suite: 0–18.0; median 0.02, average 0.17). The trace-element patterns of apatite from both groups display smooth patterns with relative enrichments of LREEs relative to HREEs. All samples, except the Yao'an apatite, show negative Eu anomalies (Online Material1 Fig. OM4).

In situ Sr isotope compositions of apatite

In situ Sr-isotope analyses (205 measurements) have been carried out on apatites from the fertile rocks including Sungun, Sarcheshmeh, Masjed Daghi, Zhunuo, Jiama, Qulong, Machangqing, and Yaoan porphyres, and on apatite grains within the barren Zhada, Nanmuqie, Renduoxiang, Wolong, and Song-gui plutons. Apatite in the fertile suites show lower radiogenic 87Sr/86Srt (0.704–0.709, median 0.706, average 0.706), whereas apatite from magmatic rocks of the barren suites has generally more radiogenic signatures (87Sr/86Srt = 0.705–0.712, median 0.708, average 0.708; Figs. 2d2f).

Apatite trace elements reveal the hydrous source of porphyry systems

The euhedral apatites crystallized early, essentially coeval with, or before monazite and xenotime (Suzuki et al. 1992). Thus, euhedral apatite can be used as an indicator mineral to trace the composition of its parental melts. The role of high contents of magmatic water during the formation of porphyry-forming magmas has been widely discussed and is recognized as a key factor in the fertility of porphyry systems (Loucks 2014; Chiaradia 2014). Apatite in ore-forming magma has high V/Y (Fig. 2b). These higher V contents in the host magma indicates elevated contents of dissolved H2O in the melt, which cause amphibole to advance in the crystallization sequence rather than titanomagnetite. V4+ partitions more strongly into magnetite than into amphibole [DVmag/melt > 130; DVamp/melt: 6.34–10.64 (La Tourrette et al. 1991; Nandedkar et al. 2016)]. Thus, at high contents of H2O in the melt, amphibole will reach saturation before titanomagnetite, so that more V remains in the residual melt, while Y continues to be efficiently extracted from the melt into amphibole [the amphibole/melt partition coefficient for Y is around 2–6 in andesitic melts (Ewart and Griffin 1994)]. Therefore, apatite trace-element proxies for amphibole in the differentiation series are especially useful indicators of the hydration state of the melt as well as the pressure at which differentiation took place (Loucks 2014; Davidson et al. 2007). This is consistent with the geochemistry of such magmas, which are characterized by small negative Eu anomalies and high V/Sc and Sr/Y (Loucks 2014), reflecting early amphibole fractionation and suppression of plagioclase crystallization in hydrous melts (Davidson et al. 2007). Moreover, the rare-earth elements entrained by apatite are incompatible during magma evolution, and thus higher REEs in apatite also may suggest a hydrous magma. In detail, LREEs (e.g., La, Ce) are more compatible in hydrous melts than HREEs and Pb. As a consequence, hydrous melts usually contain higher LREE, which can be expressed by their (Ce/Pb)N ratio (Davidson et al. 2007). It is thus expected that the apatite in hydrous magmas should have higher (Ce/Pb)N than those in the relatively dry suites. Therefore, on plots of V/Y vs. REE+Y (Fig. 2b) and V/Y vs. (Ce/Pb)N and V/Y vs. REE+Y (Fig. DR3), the fertile suites mostly have higher ratios than barren suites, suggesting early amphibole fractionation in melts and reflecting the more hydrous nature of ore-forming magmas.

Contrasting volatiles and Sr isotope compositions between fertile and barren melts in post-subduction settings

The composition of apatite can be used to calculate chlorine and sulfur concentrations in the parental melt, provided that the apatite and melt were in equilibrium, and that the experimental partition coefficients of these volatile components between the apatite and concomitant melt are known. The experimental and analytical results indicate that F, Cl, and OH in apatite can diffuse relatively rapidly along the crystallographic c-axis, and compositional gradients in these components can be observed if post-magmatic changes in temperature, pressure, and fluid or melt composition have occurred (Brenan 1993). Given these diffusion rates, the resistance of apatite to late re-equilibration may only be satisfied when the system cools rapidly (1 °C/100 yr; Brenan et al. 1995), conditions which are likely in eruptive magmas or very shallowly emplaced melts (Piccoli and Candela 2002). However, these conditions may not easily be reached in porphyry magmas. We have analyzed the apatites from core to rim by electron microprobe analysis (EMPA), to quantify the degree of diffusion (Online Material1 Fig. OM3). Cathodoluminescence and backscattered electron images were used to select analytical points. The Cl contents from core to rim show little sign of zoning, indicating our samples were not affected by diffusion. Therefore, the chlorine contents of these apatites can be used to calculate the chlorine concentrations of the coexisting melt. Several previous experimental studies have determined the partitioning behavior of Cl and/or F in the apatite-fluids-silicate melt system at different shallow-crustal conditions (Parat and Holtz 2005). We have used the recently calibrated apatite/melt chlorine partitioning model of Li and Hermann (2017) based on experimental data from Webster et al. (2009) at 200 MPa and 900 °C. We argue that the experimental conditions are reasonably close to those inferred at the time of apatite crystallization and that the thermodynamic parameters can be applied to determine the chlorine melt content in equilibrium with apatite; the detailed function and results are given in the Online Material1. Using a set of natural and experimental apatite/melt partitioning data for andesitic to rhyolitic melts, Parat et al. (2011) calibrated the following empirical non-Henrian partitioning relationship for sulfur between apatite and melt: Sapatite(wt%) = 0.0629 × ln Smelt(wt%) + 0.4513.

The application of these calculations (Table 1) shows that fertile suites have higher Cl and S contents in the melt (Cl: 0–4.8 wt%; median 0.13 wt%, average 0.22 wt%; S: 8–3490 ppm; median 19 ppm, average 64 ppm), compared with barren suites (Cl: <0.33 wt%, median 0.001 wt%, average 0.01 wt%; S: <816 ppm, median 12 ppm, average 22 ppm). For example, the parental melt of the world-class Sarcheshmeh porphyry Cu deposit contained up to 3.8 wt% Cl and 14–28 ppm S. Comparative values are 0 to 3.2 wt% Cl and 10–395 ppm S in the Qulong porphyry Cu-Mo deposit, 0.03 to 0.9 wt% Cl and 9–86 ppm S in the Jiama porphyryskarn Cu-Mo-Au deposit, and 0–4.8 wt% Cl and 15–827 ppm S in the Machangqing porphyry deposit (Table 1). These results are also similar with subduction-related porphyries system, the Coroccohuayco Fe-Cu-Au porphyry-skarn deposit, Peru (Chelle-Michou and Chiaradia 2017), and Red Chris porphyry Cu-Au deposit in British Columbia (Zhu et al. 2018). This is consistent with other suggestions that high-Cl concentrations in the melts may be a prerequisite for the formation of porphyry-related ore deposits (Hsu et al. 2019). From fluid inclusion studies, the porphyry mineralization has shown that the hypersaline liquid and vapor are enriched in chlorine sulfur (Sillitoe 2010), which are consistent with the elevated chlorine and sulfur in magmatic apatite of fertile porphyries. It is commonly accepted that chlorine and sulfur are of paramount importance for supporting the transport and deposition of ore metals at magmatic-hydrothermal systems. Specifically, chlorine and sulfur are critical elements of ore-forming fluids due to their capacity to form ligands with ore metals such as Cu, Au, Pb, Zn, Fe, Mo, as well as, with other cations such as H+, K+, Na+, and Ca+, which permit their transport to the site of ore deposition and cause hydrothermal alteration (e.g., Chelle-Michou and Chiaradia 2017).

The 87Sr/86Sr ratio in apatite can record the initial Sr isotopic value of the magma, because Rb is highly incompatible in apatite (DRbap/melt = 0.0013; Prowatke and Klemme 2006). The 87Sr/86Sr ratios of apatites are similar in these whole-rock samples (e.g., Hou et al. 2015; Lu et al. 2015). The apatites in the fertile suites show a mantle 87Sr/86Sr signature (0.704–0.706) and higher Cl/F, Cl, and S in the melts compared with barren plutons (Figs. 2d2f), suggesting that the mantle source of the magmas also supplied the volatiles necessary for mineralization. The mantle origin of apatite in the fertile suites is further supported by apatite trace-element ratios (La/Sm and Th/U) consistent with previous findings of host-rock petrogenesis worldwide (Belousova et al. 2002; Figs. 2d and 2e).

The recycled mantle origins of volatiles in the post-subduction porphyry deposits

The ore-forming magmas have higher Cl and S concentrations with less radiogenic 87Sr/86Sr (Figs. 2e and 2f). We, therefore, use these Sr isotopes and Cl content modeled mixing between altered ocean crust (AOC; mean 87Sr/86Sr 0.705; Sr 180 ppm; Staudigel et al. 1995; Cl 2000 ppm; Kendrick et al. 2017), post-subduction lithospheric mantle [PLM; Tibetan peridotite xenoliths with 87Sr/86Sr, 0.714; Sr, 83.5 ppm (Xu et al. 2017); Cl: 200 ppm], and depleted MORB mantle (DMM; 87Sr/86Sr, 0.7025; Sr, 15 ppm; Cl, 2 ppm (John et al. 2010)] to determine the sources of these volatiles. In this modeling, the chlorine from fertile porphyries requires almost 80% Cl contribution from AOC, with 10% post-subduction lithospheric mantle and/or 10% DMM (Fig. 2e). In contrast, apatite from barren suites inherits only 10–20% Cl from AOC (Fig. 2e). The plot of 87Sr/86Sr vs. apatite S (Fig. 2f) also shows a large contribution from both AOC and DMM. This modeling suggests that the higher Cl and S contents of fertile porphyries are consistent with derivation primarily from dehydration of altered oceanic basalts during subduction. This is consistent with the high-chlorine concentration of serpentinized lithosphere (up to 2000 ppm Cl; Kendrick et al. 2017) and the positive δ34S compositions of arc magmas, consistent with those of seafloor sediments (de Hoog et al. 2001). Glass inclusions in olivine from primitive arc magmas have S concentrations up to 2900 ppm (de Hoog et al. 2001), and experimental A concentrations of oxidized arc basalts can be up to 1.5 wt% S (Jugo et al. 2005). These high-chlorine and high-sulfur contents have significance for the behavior of chalcophile and siderophile metals in the genesis of porphyry Cu-Au deposit. The Cenozoic porphyry systems of Iran, Tibet, and western China developed in post-subduction settings when the AOC and DMM sources were no longer available. We thus propose that the higher chlorine and sulfur values in post-subduction Cu-Au deposits were derived from the SCLM metasomatized by previous oceanic subduction (Fig. 3).

In this model, the dehydration of altered oceanic slab (200–65 Ma; the period of during oceanic subduction) introduced a large amount of volatiles including Cl, S, and water into the SCLM of the mantle wedge (stage A; Fig. 3a), resulting in the formation of volatile-rich reservoirs in the SCLM. These may be documented by the presence of metasomatized phlogopite-bearing peridotite xenoliths entrained by Tibetan Miocene mantle-derived K-rich melts (phlogopite with 0.08 wt% Cl; Liu et al. 2011), hornblendeand phlogopite-bearing gabbro (Xu et al. 2019) in the Gangdese belts, and phlogopite- and apatite-bearing peridotites in NW Iran (Pang et al. 2013). During the continental collision that followed the subduction (stage B: ~15 Ma in Iran and Tibet; Fig. 3b), the existing volatile-bearing SCLM partially melted, with heat provided mainly by asthenospheric upwelling due to slab breakoff of the tearing or thinning of the passive continental lithospheric root (e.g., Xu et al. 2017), producing Cl- and S-enriched mantle-derived magmas (stage B; Fig. 3b). This generated widespread potassiumrich melts such as the Miocene Kerman, Miocene Gangdese, and Yunnan Eocene lamprophyres (Xu et al. 2017; Pang et al. 2013). These mafic melts underplated at the base of the crust, resulting in thermal and mass exchange during magma mixing and mingling with the overlying lower crust. This interaction led to the enrichment of volatiles in the lower crust, lowered melting temperatures to 700–750 °C (Lu et al. 2015) and promoted melting of mafic lower crust, and generated the shoshonitic and high-K calc-alkaline ore-forming porphyries (Yang et al. 2015).

The high-chlorine and high-sulfur contents of the magmas suppress the segregation of significant amounts of sulfide phases, leading to substantial metal enrichment in the evolving magma rather than in the cumulate rocks of the continental root. Thus, the degree of metal enrichment in the post-subduction settings, and the generation of world-class porphyry systems in such settings, depends on the availability of recycled chlorine and sulfur. Refertilization of the subcontinental lithospheric mantle by volatiles derived from earlier subduction is a key factor in forming post-subduction porphyries; this provides an alternative view to current ore enrichment models focusing on lower crust melting.

Porphyry systems

Relatively little geological work has been done in some areas of the Tethyan belt, such as Iran, Pakistan, Afghanistan, and even Tibet, and therefore the volatiles, trace-element contents, and Sr isotopes of apatite could usefully fingerprint fertile intrusions associated with mineralization. The robust trace-element ratios and Sr isotopes of apatite can record more information, especially of Cl and S evolution, which are key to the genesis of porphyry systems. Analyzing the compositions of apatite from an area with little geologic information or poor outcrop could efficiently and cheaply discriminate whether the drainage source area is dominated by unlikely prospective crustally derived granitoids or by prospective mantle-associated granitoids. This could help to focus exploration on the most prospective areas.

Large porphyry ore deposits can be formed by combinations of “common” geological processes, but all of these processes must be optimized to form a giant deposit. The data presented here suggest that higher volatile contents are advantageous for forming large porphyry deposits because of the optimization of ore-metal transport in such chlorine- and sulfur-rich fluids

Concluding remarks

Robust trace-element and Sr isotope data on apatite can distinguish fertile (porphyry Cu ± Au) magmas from barren magmatic suites. The fertility indicators are Cl/F (>0.19), V/Y (>0.008), and (Ce/Pb)N ratios (>2138), which are higher than in apatite from the barren suites; these elevated ratios reflect a volatile-enriched magma. Their less radiogenic Sr isotopes indicate that the higher volatile contents advantageous for ore-metal transport such chlorine- and sulfur-rich fluids reflect the derivation from the previously metasomatized mantle. This observation indicates the importance of volatiles, recycled from previous oceanic subduction, in collisional settings.

This research was funded by National Key Technologies R&D Program 2019YFA0708602, 2016YFC0600310, National Natural Science Foundation of China (42073038, 41803045), Young Talent Support Project of CAST, Fundamental Research Funds for the Central Universities and IGCP-662. The analytical data were obtained using instrumentation funded by DEST Systemic Infrastructure Grants, ARC LIEF, NCRIS/AuScope, industry partners and Macquarie University. We thank the editor Hongwu Xu, associate editor Fang-Zhen Teng and constructive comments from two reviewers, which helped in improving our paper. This is contribution 1541 from the ARC Centre of Excellence for Core to Crust Fluid Systems ( and 1409 in the GEMOC Key Centre (

Deposit item AM-21-47714, Online Material. Deposit items are free to all readers and found on the MSA website, via the specific issue's Table of Contents (go to
Open access: Article available to all readers online.