Mercury isotope evidence for the importance of recycled fluids in collisional ore systems

Abstract

The sources of fluids and metals in porphyry systems of continental-collision settings are poorly constrained. Mercury isotopes display unique mass-independent fractionation (expressed as Δ¹⁹⁹Hg) and may provide important constraints on metal and volatile sources given that Hg is a highly volatile metal. Here, we report Hg isotope data on ore-forming porphyries, barren magmatic rocks, and mantle-derived mafic magmas from southern Tibet. The fertile porphyries and coeval mafic magmas display mainly positive Δ¹⁹⁹Hg values (up to +0.25 per mil), while Δ¹⁹⁹Hg values in barren magmatic rocks and mafic magmas are largely negative (−0.54 to 0.00 per mil). The positive Δ¹⁹⁹Hg values observed here are consistent with seawater and marine sediments, suggesting that the ultimate source of fluids involved in the genesis of post-subduction porphyry copper deposits was the mantle lithosphere metasomatized by previous oceanic plate subduction. Our Hg isotope data provide an alternative view to current metallogenetic models on collisional porphyry systems that focus on melting of the lower continental crust.

Mercury isotopes provide important insights into volatile-enrichment models in collisional porphyry Cu deposits.

INTRODUCTION

Porphyry copper deposits (PCDs), which supply ~75% of Earth’s Cu and ~50% of its Mo, are formed mostly in magmatic arcs associated with syn-subduction Cl- and S-rich hydrous magmas (1–3). The volatile components act as ligands to form metal complexes, exerting a strong control on metal migration and enrichment [e.g., (4, 5)]. In subduction settings, volatiles of PCDs mainly originated from subducted oceanic slab and concentrated during intracrustal differentiation (6–8). It has been recently recognized that PCDs also can form in post-subduction settings such as those in the Alpine-Himalayan orogenic belt [e.g., (9–12)]. However, in the absence of a direct relationship to contemporaneous oceanic subduction, the origin of volatiles in post-subduction PCDs is still not well constrained, limiting our understanding of the formation of such PCDs and, therefore, mineral exploration in this geodynamic setting. The dominant view is that H₂O, Cl, and S are acquired by fertile magmas through partial melting of lower crustal rocks (13–15). However, recently, an important role for mantle-derived ultrapotassic mafic magmas has also been proposed (11, 13, 16). The Miocene Gangdese metallogenic belt in southern Tibet, a part of the Indo-Asian continental-collisional zone, extends over a length of ~550 km and a width of 100 km and hosts more than 10 large to giant post-subduction PCDs, which are natural laboratories to address the issue of the origin of volatiles in such PCDs.

Mercury (Hg) is a highly volatile metal that is more or less abundant in various types of magmatic-hydrothermal deposits. Natural Hg isotopes (¹⁹⁶Hg, ^198-202Hg, and ²⁰⁴Hg) display both mass-dependent fractionation (MDF; typically reported as δ²⁰²Hg) and mass-independent fractionation (MIF; typically reported as Δ¹⁹⁹Hg) (17). Whereas the MDF of Hg isotopes is generated via various physical, chemical, and biological processes, the MIF of Hg isotopes is generated mainly by Hg(II) photoreduction in the atmosphere-ocean system (18). Therefore, the MIF of Hg isotopes reflects Earth’s surface processes and produces negative Δ¹⁹⁹Hg values in emerged (terrestrial) environments and positive values in marine sediments and seawater (18). As expected, the primitive mantle has near-zero Δ¹⁹⁹Hg values of 0 ± 0.1 per mil (‰) (2SD) (19). Recent studies have reported both negative and positive Δ¹⁹⁹Hg in mantle-derived magmatic rocks from various geological settings, providing evidence that Hg from surface reservoirs can be recycled into Earth’s mantle via plate tectonics (20–22). MIF of Hg isotopes applied to magmatic rocks is, therefore, a powerful tool to track the recycling of volatile elements from Earth’s surface marine and terrestrial systems into the mantle and to address the question of the origin of volatile elements in post-subduction PCDs.

In this study, we analyzed Hg and Sr-Nd isotopes in 96 igneous rocks from the Gangdese belt (Lhasa block), the world’s largest post-subduction metallogenic belt [total > 45 million tonnes (Mt) of Cu; Fig. 1 and data S1] (23). The samples include 44 ore-forming porphyries from 10 PCDs, 16 barren rocks, and 36 mantle-derived mafic rocks that are temporally and spatially related to the PCDs.

Fig. 1. Topographic-geological map of the Tibetan collisional orogen.

(A) Simplified map showing tectonic outline of the Tibetan Plateau. (B) Simplified geological map of southern Tibet showing the main tectonic units and the spatial distributions of Miocene intermediate-felsic volcanic rocks, ultrapotassic mafic rocks, and coeval Gangdese porphyry deposits. V-S, volcano-sedimentary strata; BNSZ, Bangong-Nujiang suture zone; IYZSZ, Indus–Yarlung Zangbo suture zone; JSSZ, Jinshajiang suture zone; LMF, Luobadui-Milashan fault; SNMZ, Shiquan River–Nam Tso mélange zone; NL, northern Lhasa subterrane; CL, central Lhasa subterrane; SL, southern Lhasa subterrane.

RESULTS

Geological background and samples

The Tibetan Plateau consists of three continental terranes, i.e., from south to north, the Lhasa, Qiangtang, and Songpan-Ganze blocks (24). The Lhasa block is situated at the front of the collisional zone; it has a Precambrian basement in the center [as old as 2450 million years (Ma)] (25) and Phanerozoic juvenile materials added to the northern and southeastern edges (23). Extensive magmatism, driven by Neo-Tethyan oceanic subduction, occurred in the southeastern part of the Lhasa block during Triassic-Cenozoic times. During the Cenozoic, collision between the Indian and Eurasian blocks generated a series of Miocene porphyry Cu-Mo deposits (20 to 11 Ma), in the Gangdese metallogenic belt, which extends between 87° and 94° East and 29° and 30° North. Some of the most economic deposits are Qulong, Zhunuo, and Bairong (23, 24) with a total Cu endowment of >45 Mt of Cu (Fig. 1; see data S1 for description of the porphyry deposits). Ore-bearing intrusions commonly occur as isolated stocks of porphyries or monzogranite in the eastern part of the Gangdese belt (87° to 94° East). Barren intermediate-felsic rocks (24 to 13 Ma; data S1) crop out only in the western part of the belt between 80° and 87° East [e.g., (26)]. Ultrapotassic mafic rocks, including basalts, trachy-basalts, and lamprophyres with K₂O > 3 wt % and MgO >3 wt %, are widespread throughout the whole belt. The eastern ultrapotassic mafic rocks (EUMRs) have ages of 12 to 9 Ma and crop out as dikes in the Gangdese porphyry Cu deposits (e.g., Bairong, Zhunu, and Jiama). As an alternative to partial melting of the lower crust (13–15), mixing of the EUMRs with partial melts of the lower crust has been suggested to have promoted the formation of ore-forming porphyries by providing exogenous water (11, 16). The western ultrapotassic mafic rocks (WUMRs) are 18- to 11-Ma old (Fig. 1 and data S1) and occur as lava flows unconformably overlying Cretaceous-Paleogene volcanic-sedimentary sequences [e.g., (27)]. The samples studied here comprise ore-bearing porphyries in the east, barren rocks in the west, and mantle-derived ultrapotassic mafic rocks along the whole Gangdese metallogenic belt.

Hg concentrations and isotope composition

Total Hg (THg) concentrations and isotopic compositions of the samples are given in data S2. The fertile porphyries display overall low THg concentrations [0.4 to 20.6 parts per billion (ppb)], and large variations in δ²⁰²Hg (−3.08 to 0.24‰) and ∆¹⁹⁹Hg (−0.25 to 0.22‰ with a median of −0.01 ‰) values. The coeval EUMRs have THg concentrations of 0.6 to 1.9 ppb, δ²⁰²Hg of −3.34 to −0.40‰, and ∆¹⁹⁹Hg of −0.24 to 0.25‰ with a median of 0.05 ‰. The barren felsic rocks in the western part are characterized by low THg concentrations (0.8 to 1.2 ppb) and negative values of δ²⁰²Hg (−2.06 to −0.17‰) and ∆¹⁹⁹Hg (−0.30 to 0.00‰ with a median of −0.18‰), similar to coeval WUMRs that have δ²⁰²Hg of −2.78 to −0.51‰ and ∆¹⁹⁹Hg of −0.54 to −0.15‰ with a median of −0.33‰. Only fresh samples were selected for geochemical and isotopic analyses (see representative photomicrographs in fig. S1), and the loss on ignition (LOI) do not display correlations with THg, δ²⁰²Hg, or ∆¹⁹⁹Hg (fig. S2), further indicating that alteration has had no influence on Hg isotopes.

DISCUSSION

Hg-MIF signature of oceanic subduction in fertile porphyries and coeval mantle rocks

Many studies suggest that magmatic, metamorphic, and hydrothermal processes do not affect Hg-MIF (28–30). The lack of correlations between SiO₂ concentrations and δ²⁰²Hg or ∆¹⁹⁹Hg values (Fig. 2, A to C) indicates that Hg isotope variations are not related to magma differentiation processes. In contrast, all samples show positive correlations between ∆¹⁹⁹Hg and ∆²⁰¹Hg with a slope of 1.02 ± 0.12 (2 SE; fig. S3), identical within uncertainty to the correlation associated with aqueous Hg(II) photoreduction (slope of 1.00 ± 0.02) (17). Therefore, the Hg-MIF signals observed in porphyries and coeval mantle-derived rocks are not affected by magmatic processes, and these signals are diagnostic of the recycling of surface Hg into the mantle (19, 21, 28). Now, there is limited research on the effect of degassing on Hg-MIF. Zambardi et al. (30) did not observe Hg-MIF in volcanic fumaroles. In contrast, a weak correlation between Δ¹⁹⁹Hg and δ²⁰²Hg with a slope of −0.023 in terrestrial igneous rocks has been attributed to magma degassing (31). Nonetheless, our data do not show such a correlation (Fig. 3).

Fig. 2. Covariations of Hg-MIF and Hg-MDF with SiO₂ and longitude.

(A and B) ∆¹⁹⁹Hg versus SiO₂ and longitude. (C and D) δ²⁰²Hg versus SiO₂ and longitude.

Fig. 3. Mercury isotopic compositions of the Miocene rocks in the Gangdese belt.

(A) Plots of Δ¹⁹⁹Hg versus Δ²⁰¹Hg and (B) δ²⁰²Hg versus Δ¹⁹⁹Hg. Fields for marine reservoir, terrestrial reservoir, and primitive mantle are from (21). Box plots are shown to demonstrate the distribution of Hg isotope values; the 25 to 75% range (solid bars), range within 1.5 interquartile range (whiskers), median (black line in the bars), and outliers (solid circles).

The ultrapotassic mafic rocks from the Lhasa block are interpreted to be derived from partial melting of the slab-metasomatized lithospheric mantle [e.g., (16, 32–34)]. This is supported by their high Cr [62 to 741 parts per million (ppm)] and Ni (32 to 588 ppm) contents (data S2), high abundances of large ion lithophile elements, low abundances of high field strength elements (fig. S4), and enriched Sr-Nd isotopic signatures (data S2). The occurrence of nonzero Δ¹⁹⁹Hg values (−0.54 to 0.25‰) in the Tibetan mantle-derived ultrapotassic mafic rocks, in contrast to estimates for the primitive mantle value (0 ± 0.1‰, 2SD) (19), is consistent with recent results on magmatic rocks interpreted as derived from slab-metasomatized mantle (19, 22, 35). Pronounced Hg-MIF signals in the metasomatized mantle, mainly positive Δ¹⁹⁹Hg values, have been explained by the addition of marine Hg via the subduction of marine sediments with the oceanic slab, because marine materials show mainly positive Δ¹⁹⁹Hg values (19, 21, 22).

An alternative possibility to explain the occurrence of Hg-MIF signatures in the EUMRs could be magmatic assimilation of marine (meta)sediments that have positive Δ¹⁹⁹Hg values and tectonically emplaced into the lower crust. However, heating and assimilation of metasedimentary rocks by mantle-derived melts should result in the formation of coeval S-type melts [e.g., (36)], which are not observed in eastern Gangdese. In addition, in contrast to the WUMRs (37), crustal xenoliths are absent in the EUMRs. Compared to ultrapotassic rocks from the Alban Hills (Roman magmatic province), which assimilated limestone or dolostone (38), EUMRs exhibit lower and narrower CaO/SiO₂ values (<0.2; fig. S5). These lines of evidence suggest that the EUMRs have not assimilated marine metasediments.

We suggest that the mostly positive Δ¹⁹⁹Hg values of EUMRs (Fig. 2, A and B) and their topology in the Δ²⁰¹Hg versus Δ¹⁹⁹Hg and Δ¹⁹⁹Hg versus δ²⁰²Hg diagrams (Fig. 3), overlapping with data from marine reservoirs (21), indicate the occurrence of subducted recycled marine materials in the mantle source of EUMRs. This is also supported by their Sr-Nd isotopic compositions [[⁸⁷Sr/⁸⁶Sr]_i = 0.70741 to 0.72764 and ε_Nd(t) = −14.8 to −5.5; data S2], which overlap with those of the oceanic sediments (16), and by Hf-Nd isotopic features close to the seawater array, with low Hf/Sm ratios similar to pelagic sediments and marine pelites [(16) and the references therein]. Multiple stages of Tethyan oceanic subduction occurred beneath the Tibetan region since the Paleozoic (39, 40), and recycling of marine sediments in the Tibetan ultrapotassic rocks has been supported by other recent isotopic data, i.e., isotopically light Mo, Mg, and K (δ^98/95Mo = −0.45 to −0.13‰, δ²⁶Mg = −0.46 to −0.22‰, and δ⁴¹K = −1.55 to −0.32‰) (33, 34, 41).

In contrast to EUMRs, the WUMRs have the lowest Δ¹⁹⁹Hg values (as low as −0.54‰), overlapping with the field of the terrestrial reservoir (Fig. 3) (18). Granitic rocks and metasedimentary enclaves that derived from upper continental crust have small Δ¹⁹⁹Hg variations of −0.16 to 0.09‰ with a weighted average value of 0.03 ± 0.15‰ (42), which is higher than that of the WUMRs (−0.54 to −0.15‰). Although Hg isotope measurements for lower crustal rocks are not now available, the relatively lower ⁸⁷Sr/⁸⁶Sr ratios of lower crustal xenoliths (~0.708) than that of the host WUMRs (0.711 to 0.722) suggest that the enriched nature of the WUMRs is inherited from their mantle source (37). Similar negative Δ¹⁹⁹Hg values (−0.45) were also observed in Pitcairn basalts, suggesting recycling of ancient terrestrial Hg into the mantle (19). Recycling of blueschist facies mélange from passive continental margin during plate convergence could make up a mantle source of post-collisional potassic and ultrapotassic rocks (43). The negative Hg-MIF signatures of terrestrial materials could also be transferred to the mantle lithosphere by a similar mechanism.

Overall, the results of this study confirm that the lithospheric mantle of the Lhasa block has experienced different degrees and cycles of subduction-related fertilization with higher proportions of marine input in the eastern region and of terrestrial input in the western region. Precambrian to early Paleozoic mafic magmatic rocks, such as ~570-Ma gabbros (εNd = −17.6 to −11.0) (44) and ~490-Ma metamorphic volcanic rocks (⁸⁷Sr/⁸⁶Sr = 0.73 to 0.77 and εNd = −7.5 to 0.3) (45), reveal that the Mozambique and Proto-Tethys oceanic slab have been subducted beneath the Lhasa block. The subsequent subduction of the Bangong-Nujiang and Indus–Yarlung Zangbo Tethyan oceanic slab could have further modified the lithospheric mantle of the Lhasa block (39). Therefore, repeated subduction processes have caused different degrees of modification of the lithospheric mantle by oceanic and ancient terrestrial sediments on both sides of the Gangdese belt, leading to the differences in the systematics of Hg isotopes with mostly negative Δ¹⁹⁹Hg values in the east and mostly Δ¹⁹⁹Hg positive values in the west.

There are two main hypotheses for the origin of the Miocene Cu-rich magmas in the Gangdese belt: (i) partial melting of a subducted oceanic slab (46, 47) or (ii) mixing between partial melts of a juvenile lower crust (48) and metasomatized mantle-derived ultrapotassic magmas (11, 16). Our data show that the fertile porphyries have higher Δ¹⁹⁹Hg values (−0.22 to 0.22‰) than those of the oceanic crust (0.0 to 0.2‰) (49) and higher ⁸⁷Sr/⁸⁶Sr ratios (0.7049 to 0.7099) than the Yarlung-Zangbo Neo-Tethyan ophiolite (0.7030 to 0.7047) (50), making it unlikely that slab melts are the main source of the fertile magmas. In contrast, our data are consistent with an origin of fertile magmas by mixing between lower crustal partial melts and ultrapotassic magmas as discussed below.

The similar, mostly positive, Δ¹⁹⁹Hg signatures of the fertile porphyries and coeval mafic ultrapotassic magmas point to the same source of Hg and, by inference, of other volatiles in both these rocks. The suggested source of Hg is subducted oceanic sediments based on their unique positive Δ¹⁹⁹Hg. The high average Hg content in marine sediments (~177 ppb with a peak at 837 ppb) (51, 52), 100× higher than that in mid-ocean ridge basalt (MORB) [~1.36 ppb; (21)] and substantially higher than that in the continental crust (~8 to 50 ppb) (42, 53), allows for the efficient modification of the overlying mantle during recycling of marine sediments via subduction. This process is supported by the positive Δ¹⁹⁹Hg values of arc magmas (−0.01 to 0.34‰) (21), similar to those of seafloor sediments [−0.15 to 0.45‰; (51, 52)]. A Monte Carlo mixing model between the depleted MORB mantle (DMM) and oceanic sediments shows that involvement of less than 5% of oceanic sediment–derived fluids in the mantle source could explain most of the Sr-Nd-Hg isotope features of the eastern ultrapotassic rocks and porphyries (Fig. 4).

Fig. 4. Monte Carlo simulation for mixing between mantle and oceanic components.

(A) Plots of δ²⁰² Hg versus Δ¹⁹⁹Hg, (B) Δ²⁰¹Hg versus Δ¹⁹⁹Hg, (C) Δ¹⁹⁹Hg versus ⁸⁷Sr/⁸⁶Sr, and (D) Δ¹⁹⁹Hg versus ¹⁴³Nd/¹⁴⁴Nd. The small dots with different colors represent random mixing of fluids derived from oceanic components with the depleted MORB mantle (DMM) at variable proportions from a Monte Carlo simulation. DMM: The contents of Hg (0.5 ppb), Sr (7.66 ppm), and Nd (0.58 ppm) are from (91, 92), and the Hg-Sr-Nd isotopic compositions are from (19, 93). The Hg isotopes of oceanic components are represented by oceanic sediments (51, 55, 76–86) and marine sedimentary rocks (87, 88). Their Sr-Nd elemental compositions are from global subducting sediment (GLOSS) (94), and isotopic compositions (⁸⁷Sr/⁸⁶Sr = 0.73631 and ¹⁴³Nd/¹⁴⁴Nd = 0.51186) are from marine sediments (Java and Cascades trenches) that were reported by (95, 96).

The Hg isotope data for adakitic porphyries show elevated Δ¹⁹⁹Hg values toward EUMRs (Figs. 3 and 4), in line with a mixing process, and this is strengthened by the findings of high-Mg diorites in the Qulong and Zhunuo PCDs that are interpreted as mixing products of adakite-like and mafic ultrapotassic rocks (11, 16). Injection into the lower crust of mafic magmas derived from metasomatized mantle could lower the crustal melting temperature to 700° to 750°C (54). This process promotes the formation of volatile-rich (e.g., Hg, Cl, H₂O, and S) ore-forming magmas (11) and results in positive Hg MIF values (Fig. 3). In the western Gangdese belt, Miocene PCDs are rare. The negative Δ¹⁹⁹Hg values observed in the mantle-derived rocks of the western Gangdese suggest a minor contribution from oceanic components, further underscoring the importance of earlier oceanic subduction and associated recycling of volatiles for the formation of porphyry Cu deposits in continental collision zones (55).

Although our Hg isotope data indicate a primary difference in the supply modality of volatiles between the barren western and fertile eastern part of the Gangdese belt, we cannot exclude that other processes were also responsible for the different fertility of magmas in the west and east of the Gangdese belt. For instance, the Miocene felsic rocks in the western Gangdese were derived from ancient crustal sources and usually consist of a single intrusive phase with an overall short lifetime. In contrast, the fertile porphyries in the eastern Gangdese were formed from partial melting of juvenile lower crust and usually consist of multiple long-lived magmatic pulses (56, 57), which is a typical feature of magmatism associated with supergiant PCDs (8, 11, 58). Nonetheless, our Hg isotope data allow us to suggest that melts that acquired fluids from a slab-metasomatized mantle played a major metallogenetic role in the formation of post-subduction porphyry Cu deposits compared to melts derived from partial melting of the lower crust.

Implications for the recycled mantle-derived volatiles in post-subduction porphyry deposits

High magmatic water content is a typical feature of fertile magmas associated with porphyry deposit formation (58–60). Long-lasting injection of hydrous mafic melts in the lower crust leads to the formation of large amounts of andesitic magma with high H₂O concentrations (5.5 to 13 wt %) (58), most of which (80 to 95%) is H₂O from residual mantle-derived melts, with only <5 to 20% being H₂O associated with dehydration melting of amphibolite (61). Our results further indicate that the contribution of hydrous melts from a mantle metasomatized by volatiles derived from oceanic sediment subduction is essential for the formation of post-subduction PCDs in the east belt. Experimental studies also suggest that partial melts of amphibolite at 1.5 GPa and 800° to 950°C are felsic with MgO < 2 wt % (62), lower than most Gangdese porphyries (0.1 to 4.4 wt %; data S2). The input of mantle-derived shoshonitic mafic melts to amphibolite melts can provide sufficiently high concentrations of incompatible elements characteristic of adakite-like potassic magmas (63). A mantle-derived volatile-rich reservoir in the continental lithospheric mantle (CLM) is also implied by the presence of metasomatized phlogopite-bearing peridotite xenoliths entrained in Tibetan Miocene mafic rocks (64), of hornblende- and phlogopite-bearing gabbro (65) in the eastern Gangdese belt, and of phlogopite- and apatite-bearing pyroxenite xenoliths in northwestern Iran (66).

Together with Hg, volatiles such as Cl and S, which control the complexing and transport of chalcophile metals in hydrothermal solutions, are also released from the subducted oceanic sediment and crust (55, 67) (stage A; Fig. 5A), as seafloor sediments have higher concentrations of chlorine (164 ppm) and sulfur (13.3 ppm) than depleted mantle (Cl = 0.17 ppm and S = 6.08 ppm) (68). During continental collision following subduction (stage B; Fig. 5B), tectono-thermal events, such as the ~20- to 10-Ma asthenosphere upwelling in the eastern Gangdese (69), induced partial melting of the slab-metasomatized CLM forming volatile-rich ultrapotassic mafic magmas, which are considered to play a primary metallogenetic role in the formation of the PCDs of the eastern Gangdese belt (11, 16). In summary, our study, highlighting that the fertilization of the CLM by volatiles derived from earlier oceanic subduction is a key factor in forming post-subduction porphyries, provides an alternative view to current metallogenetic models that focus on melting of the lower crust. It also provides additional evidence for large-scale translithospheric recycling of crustal Hg via plate tectonics.

Fig. 5. Evolution of volatiles in post-subduction settings.

(A) Stage A: During oceanic slab subduction, dehydration of the subducting oceanic crust and sediment leads to partial melting of the asthenosphere. Volatile-enriched mafic melts intrude the overlying lithosphere and enrich Cl, S, and water in the continental root, generating phlogopite-bearing CLM. (B) Stage B: During the continental collision period, upwelling of asthenospheric mantle triggers remelting of previously metasomatized volatile-enriched CLM, introducing melts capable of scavenging chlorine and sulfur from the formerly enriched lithosphere root. Underplating of these Cl- and S-rich mafic magmas induces mixing and mingling with the overlying lower crust, leading to formation of fertile magmas for post-collisional porphyry systems. The Δ¹⁹⁹Hg values of upper continental crust, asthenosphere (primitive mantle), and oceanic crust are from (42) and the references therein. The Δ¹⁹⁹Hg values of oceanic sediments are from (51, 52, 76–86).

MATERIALS AND METHODS

Chemical analyses

THg concentrations and Hg isotope ratios were analyzed at the Institute of Geochemistry, Chinese Academy of Sciences in Beijing. Before chemical analysis, the samples were powdered to 200 mesh size and homogenized in an agate mortar. THg concentration was determined by a Lumex RA-915 + Hg analyzer equipped with a PYRO-915 + attachment (Russia), with a detection limit of 0.5 ng/g. Standard reference material GSS-4 (soil) was simultaneously tested, yielding Hg recoveries of 90 to 105% and RSD of <9%.

The samples were prepared for Hg isotope analysis using a double-stage tube furnace coupled with 40% anti aqua regia (HNO₃/HCl = 3/1, v/v) trapping solutions (70).

Standard reference materials GSS-4 were prepared in the same way as the samples. The prepared samples were diluted to 0.5 ng/ml with an acid concentration of 10 to 20% before Hg isotope analysis using a Neptune Plus multi-collector inductively coupled plasma mass spectrometry (71). THg concentrations and acid matrices in the bracketing National Institute of Standards and Technology (NIST) SRM 3133 solutions were matched well with the neighboring samples. Hg isotope ratios were reported following the convention proposed by Blum and Bergquist (72). Specifically, MDF is expressed in δ²⁰²Hg notation in units of ‰ referenced to the NIST-3133 Hg standard (analyzed before and after each sample)

$$\begin{array}{c}\mathrm{\delta }{}^{\mathit{xxx}}\mathrm{H}\mathrm{g}\left(‰\right)=\left[\left({}^{\mathit{xxx}}\mathrm{H}\mathrm{g}/{}^{198}\mathrm{H}_{\mathrm{g}}^{\text{sample}}\right)/\right.\hfill \\ \left.\left({}^{\mathit{xxx}}\mathrm{H}\mathrm{g}/{}^{198}\mathrm{H}_{\mathrm{g}}^{\text{standard}}\right)-1\right]\times 1000\hfill \end{array}$$ (1)

Hg-MIF is reported in Δ notation, which describes the difference between the measured δ^xxxHg and the theoretically predicted δ^xxxHg, in units of per mil

$$\mathrm{\Delta }{}^{\mathit{xxx}}\mathrm{H}\mathrm{g}\approx \mathrm{\delta }{}^{\mathit{xxx}}\mathrm{H}\mathrm{g}-{\mathrm{\delta }}^{202}\text{Hg}\times \mathrm{\beta }$$ (2)

where xxx is 199, 200, or 201; and β is 0.2520 for ¹⁹⁹Hg, 0.5024 for ²⁰⁰Hg, and 0.7520 for ²⁰¹Hg. NIST-3177 secondary standard solutions, diluted to Hg (0.5 ng/ml) with 10% HCl, were measured every 10 samples to monitor the instrument stability.

The overall average and uncertainty of NIST-3177 (δ²⁰²Hg, −0.60 ± 0.10‰; Δ¹⁹⁹Hg, −0.01 ± 0.08‰; Δ²⁰⁰Hg, 0.01 ± 0.06‰; and Δ²⁰¹Hg, −0.03 ± 0.07‰; 2SD, n = 14) and GSS-4 (δ²⁰²Hg, −1.43 ± 0.06‰; Δ¹⁹⁹Hg, −0.42 ± 0.07‰; Δ²⁰⁰Hg, −0.02 ± 0.07‰; and Δ²⁰¹Hg, −0.36 ± 0.09‰; 2SD, n = 4) agree well with previous results (19–21). The larger of the values of SD (2 SD) for either NIST-3177 or GSS-4 are used to reflect analytical uncertainties of the samples.

Whole-rock major- and trace-element and Sr-Nd isotope analyses were conducted at Wuhan SampleSolution Analytical Technology Co. Ltd. LOI was calculated on the basis of major-element concentrations. Detailed analytical methods are provided in Supplementary Text. The Sr-Nd isotopic compositions of pure materials and USGS rock standards (BCR-2 and RGM-2) are identical within error to published values (73–75), and all results are listed in data S2.

Monte Carlo isotope mixing model

To investigate the compositional diversity of sediments across a wide range of scenarios, we randomly selected three sediments from recently reported global oceanic sediments (51, 52, 76–86) and marine sedimentary rocks (87, 88). These sediments were mixed in random proportions to create a new sediment endmember. Subsequently, fluids including aqueous fluid and hydrous melt derived from this sediment endmember were calculated assuming random dehydration in the range of 0 to 10%. The partition coefficients of Sr and Nd between sediments and fluids under varying temperature and pressure were used for modeling (89, 90). The partition coefficients of Hg (residue/fluid) were allowed to range randomly between 0.1 and 1. We assume that no Sr-Nd-Hg isotope fractionation occurred during dehydration processes, as suggested by now available data. This newly calculated sediment-derived fluid was then mixed with the mantle in random proportions (0 to 20%).

The THg content and Δ¹⁹⁹Hg values of sediment fluid (sf)–mantle mixture can be calculated using the following mass balance equations

$${\left[\text{THg}\right]}_{\text{mix}}={\left[\text{THg}\right]}_{\text{sf}}\times F+{\left[\text{THg}\right]}_{\text{mantle}}\times \left(1-F\right)$$ (3)

$$\begin{array}{c}{\left[\text{THg}\right]}_{\text{mix}}\times {\mathrm{\delta }}^{\mathit{xxx}}{\text{Hg}}_{\text{mix}}={\left[\text{THg}\right]}_{\text{sf}}\times {\mathrm{\delta }}^{\mathit{xxx}}{\text{Hg}}_{\text{sf}}\times \hfill \\ F+{\left[\text{THg}\right]}_{\text{mantle}}\times {\mathrm{\delta }}^{\mathit{xxx}}{\text{Hg}}_{\text{mantle}}\times \left(1-F\right)\hfill \end{array}$$ (4)

$$\begin{array}{c}{\left[\text{THg}\right]}_{\text{mix}}\times {\mathrm{\Delta }}^{\mathit{xxx}}{\text{Hg}}_{\text{mix}}={\left[\text{THg}\right]}_{\text{sf}}\times {\mathrm{\Delta }}^{\mathit{xxx}}{\text{Hg}}_{\text{sf}}\times \hfill \\ F+{\left[\text{THg}\right]}_{\text{mantle}}\times {\mathrm{\Delta }}^{\mathit{xxx}}{\text{Hg}}_{\text{mantle}}\times \left(1-F\right)\hfill \end{array}$$ (5)

where xxx is 199, 200, or 201; and F refers to the sediment mass flux into the mantle.

Similarly, the Sr and Nd concentrations and ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd ratios of mixture of sediment-derived fluid and mantle can also be obtained. Using this approach, we generated a total of 60,000 mixtures. The results of our modeling are shown in Fig. 4.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S5

Legends for data S1 and S2

References

Other Supplementary Material for this manuscript includes the following:

Data S1 and S2