Introduction

Seafloor massive sulfide (SMS) deposits form as part of submarine hydrothermal systems that are driven by volcanism along both divergent and convergent plate boundaries1,2,3. They are a potential future source of metals deemed critical for technology and industry (e.g., copper, zinc, and gold)4 and represent modern analogs to volcanogenic massive sulfide (VMS) deposits currently exploited in ancient volcanic successions onshore5. In addition, hydrothermal venting atop active SMS deposits creates a unique biological habitat that is argued to have supported early life on Earth6,7 and fluxes a wide range of chemicals from the oceanic crust into the oceans. In this context, it is increasingly recognized that hydrothermal vents are contributors to the oceanic inventory of dissolved iron, a key micronutrient that regulates phytoplankton growth and hence the biological carbon pump8,9,10,11,12,13.

SMS genesis is considered to involve (1) exhalative sulfide deposition, forming chimney and mound structures and vent-distal stratiform ores at the seafloor, and (2) sub-seafloor infilling and replacement of primary hydrothermal precipitates and underlying volcanic material by sulfide5,14,15,16,17,18. Sub-seafloor mineralization in particular is shown to be key in producing large-tonnage sulfide deposits, yet the nature of the different hydrothermal processes involved remains poorly constrained5,15,16,18. Since evidence for such processes is commonly obscured by geologic overprinting in fossil onshore VMS deposits, ocean drilling of active SMS systems represents a crucial source of information for detailed characterization of sub-seafloor hydrothermal activity in a range of tectonic settings18,19,20,21,22. To this end, stable iron isotopes in sulfides from the internal portions of an active SMS deposit offer a method to further elucidate sub-seafloor hydrothermal processes, including their role in the oceanic (bio-)geochemical cycling of iron23. Moreover, such in situ sub-seafloor iron isotope data have the potential to better resolve existing iron isotope interpretations for hydrothermal vents and plumes atop active SMS systems and can help reconcile them with observations from fossil onshore VMS deposits and from experimental studies of high-temperature hydrothermal sulfide formation8,13,24,25,26,27,28,29,30,31,32,33,34, which would broaden the applicability of iron isotopes as a geochemical tracer.

Here, we apply iron isotope analysis of sulfide minerals to a suite of historical drill cores from the Trans-Atlantic Geotraverse (TAG) active mound and stockwork complex of the north-central Mid-Atlantic Ridge (Fig. 1). The TAG complex formed through intermittent cycles of high-temperature hydrothermal activity over at least the last 20,000 years35,36,37 and its detailed growth history includes repeated hydrothermal precipitation at and below the mound surface, mechanical and hydrothermal reworking of earlier formed precipitates, and progressive alteration and partial incorporation of the basaltic host rocks15,19,38,39. Building on this geologic and hydrothermal framework, we use our new sulfide iron isotope data to further constrain sub-seafloor hydrothermal processes at TAG and assess their implications for ore formation and for iron cycling in the North Atlantic Ocean. The unique insights gathered from our study of TAG will likely enhance the utility of iron isotopes to the investigation of active and fossil seafloor hydrothermal systems elsewhere.

Fig. 1: Bathymetric maps with sampling locations.
figure 1

a Location of TAG and other hydrothermal vent fields and sulfide deposits (red symbols) along the TAG segment, north-central Mid-Atlantic Ridge: 1—Lost City; 2—MAR 30°N; 3—Broken Spur; 4—MAR 24°30’N; 5—Snakepit; 6—Surprise; 7—Puy des Folles; 8—Zenith-Victory; 9—Yubileinoe. Map modified from ref. 68 and ref. 69 and retrieved via GeoMapApp (http://www.geomapapp.org). b Detailed map of the TAG mound showing the collar locations of the ODP Leg 158 drill cores used in this study (957C, 957E, 957H, 957M, and 957P). Map modified from ref. 70 and retrieved via GeoMapApp.

Full size image

The TAG mound and stockwork complex

The TAG active sulfide mound was discovered in 198540 at the E margin of the rift valley of the slow-spreading Mid-Atlantic Ridge at 26°N (Fig. 1) and is located in an area featuring several inactive sulfide and oxide deposits41,42,43. Following its discovery, TAG was the first-ever SMS deposit investigated by the Ocean Drilling Program (ODP) in 199444,45 and it is now one of the World’s most comprehensively characterized modern seafloor hydrothermal systems. The mound is located about 2.5 km E of the neovolcanic zone of the ridge at ~3670 m water depth. It forms a distinctly circular structure composed of two superposed terraces, and it measures ~200 m in basal diameter and rises ~50 m above the seafloor19,46 (Fig. 1b). The mound is underlain by a ~80 m diameter, pipe-shaped stockwork of mineralized and intensely altered basaltic basement rocks that extends to at least 125 m below the surface19,38 (Fig. 2a). The mound and underlying stockwork are estimated to contain a combined mass of ~3.9 million tonnes of sulfide with an average grade of 1–2% copper. The morphology, size, and bulk composition of TAG is comparable to that of some VMS deposits preserved in ophiolites, e.g., those exploited in Cyprus, in Oman, and in Newfoundland47.

Fig. 2: Internal structure of TAG and examples of sampled sulfide textures.
figure 2

a Composite section through the TAG mound and underlying stockwork drawn based on multiple drill cores in each area; TAG-5 is projected onto the section. The samples utilized in this study (yellow diamonds) encompass the main rock types from the vertical and lateral extent of TAG19,44: massive pyrite and pyrite breccia (types 5 and 6); pyrite-anhydrite breccia (type 7) and pyrite-silica(−anhydrite) breccia (type 8), both with abundant anhydrite veins (type 11); pyrite-silica and silicified wallrock breccia (types 9 and 10a); and chloritized basalt breccia (type 10b). Vent fluid temperatures for the Black Smoker Complex and the (now inactive) Kremlin Area are from ref. 49 and isotherms are drawn based on fluid inclusion data39,71. Figure modified from ref. 52 and ref. 69. b Massive sulfide mineralization from the upper parts of the mound comprising fine-grained and porous pyrite and marcasite with local chalcopyrite. The top part of the rock piece preserves a colloform texture (TAG-4, 957M, 14.79 m.b.s.f.). c Aggregates of pyrite and chalcopyrite (partly oxidized, greenish in photo) within a fragment of an anhydrite vein from the lower part of the mound (TAG-1, 957C, 46.69 m.b.s.f.). d Pyrite-silica breccia from the upper part of the stockwork comprising clasts of massive granular pyrite in a matrix of quartz and pyrite (TAG-2, 957H, 41.32 m.b.s.f.). e Fragment of chloritized basalt host rock with finely disseminated pyrite from the deeper part of the stockwork. The altered basalt is crosscut by quartz-pyrite stringer veins (TAG-1, 957E, 116.42 m.b.s.f.).

Full size image

Presently, high-temperature (360–370 °C), copper-rich black smokers are discharged from a cluster of chalcopyrite-pyrite-anhydrite chimneys situated on top of a ~20 m diameter cone structure in the NW part of the upper terrace (the ‘Black Smoker Complex’; Figs. 1b and 2a)48,49,50. A field of sphalerite-dominated, beehive-shaped chimneys occurs on the lower terrace ~70 m SE of the Black Smoker Complex and is named the ‘Kremlin Area’ (Figs. 1b and 2a). Prior to and during ODP drilling, these chimneys vented white smokers that were distinctly colder (260–300 °C) and were copper-iron-poor, but zinc-rich relative to the black smoker discharges. These white smokers were suggested to have evolved from an end-member hydrothermal fluid via conductive cooling and mixing with entrained seawater, and precipitation of sulfides and anhydrite, within the mound48,49,50. The white smoker venting was, however, observed to have ceased during revisits in 2003 and later51.

The mound surface is made up entirely of hydrothermal precipitates including plate-like crusts and bulbous blocks of porous massive sulfide as well as siliceous iron oxyhydroxide sediments15. Oxidized pyrite-rich sulfide talus and debris flows form an apron that surrounds and partially covers the steep-sided margins of the mound48. Below the surface, the upper part of the mound (0 to ~15 m.b.s.f.) consists of massive pyrite and massive pyrite breccias with locally abundant iron oxyhydroxides, chert, chalcopyrite, sphalerite, and marcasite15,19,39,44,52 (Fig. 2a). The massive pyrite zone is underlain by an anhydrite-rich zone made up of pyrite-anhydrite and pyrite-silica(-anhydrite) breccias that extends down to ~45 m.b.s.f. (Fig. 2a). The breccias in this zone are extensively crosscut by up to ~45 cm thick, composite anhydrite veins with localized sulfide mineralization53. The pyrite-silica(-anhydrite) breccias at the base of the anhydrite-rich zone define the transition from the mound to the upper part of the stockwork, which comprises pyrite-silica breccias that grade into silicified wallrock breccias with depth39 (Fig. 2a). Below ~100 m.b.s.f., the silicified wallrock breccias eventually transition into chloritized basalt breccias that are overprinted by multiple generations of pyrite, quartz, and quartz-pyrite stringer veins19 (Fig. 2a).

Results

Iron isotope compositions were determined for a total of 50 micro-drilled sulfide separates encompassing the main rock types from the vertical and lateral extent of the TAG mound and the underlying stockwork (Fig. 2a). The TAG sulfides show an overall range of δ56Fe values between −1.27 and +0.56‰, with a mean value of +0.04‰ (±0.08‰ 2σ; Table 1, Supplementary Data 1). Notably, variations in δ56Fe values are observed between the different textural types of sulfide analyzed, including (following ref. 53): (1) massive sulfide; (2) vein-related sulfide; (3) sulfide in clasts from different breccia types; and (4) disseminated sulfide in altered basalt host rock.

Table 1 Summary of iron isotope data for TAG sulfides.
Full size table

Massive sulfide

Massive sulfide (sulfide-only) samples from the upper portions of the TAG mound are dominated by fine-grained, variably porous pyrite that in places preserves primary colloform and spheroidal textures15 (Fig. 2b). Similar variably porous massive sulfide occurs locally also as cement in silicified wallrock breccias at deeper levels. Pyrite from massive sulfide samples shows a mean δ56Fe value of −0.42‰ and a range from −1.27 to +0.28‰ (n = 9; Table 1, Supplementary Data 1).

Vein-related sulfide

The anhydrite vein networks within the pyrite-anhydrite and pyrite-silica(-anhydrite) zones contain sulfides locally. Fine-grained pyrite and chalcopyrite are aggregated in bands or clots within the veins (Fig. 2c), with more voluminous sulfide present in selvages and halos along the vein margins15,19,53. Pyrite samples from the anhydrite veins and selvages show δ56Fe values between −0.52 and +0.06‰ with a mean of −0.24‰ (n = 5), whereas co-existing chalcopyrite records δ56Fe values between −0.04 and +0.08‰ with a mean of +0.01‰ (n = 3; Table 1, Supplementary Data 1).

Coarse-grained pyrite from quartz-pyrite stringer veins crosscutting the chloritized basalt breccias in the deeper parts of the TAG stockwork (Fig. 2e) shows δ56Fe values between +0.09 and +0.38‰ with a mean of +0.22‰ (n = 3; Table 1, Supplementary Data 1).

Sulfide in breccia clasts

Clasts of massive sulfide occur throughout the pyrite, pyrite-anhydrite, pyrite-silica(-anhydrite), and the pyrite-silica breccias at TAG (Fig. 2d). The breccia clasts range from a few mm up to >5 cm in diameter and are chiefly made up of compact, granular aggregates of polycrystalline euhedral pyrite15. Some sulfide clasts, however, contain fine-grained microporous domains that preserve remains of colloform and spheroidal mineral textures, whereas some other clasts exhibit sequential growth banding15. Locally preserved microcrystalline pyrite textures are also noted and may represent a precursor to some coarser pyrite clasts15. Pyrite from the breccia clasts exhibits δ56Fe values between −0.32 and +0.37‰ with a mean of +0.12‰ (n = 22; Table 1, Supplementary Data 1).

Disseminated sulfide in altered basalt host rock

The finely disseminated pyrite samples extracted from within variably altered basalt fragments in the silicified wallrock and chloritized basalt breccias of the TAG complex (Fig. 2e) yield a mean δ56Fe value of +0.47‰ and a range from +0.34 to +0.56‰ (n = 8; Table 1, Supplementary Data 1).

Discussion

Assessment of hydrothermal fluid iron isotope fractionation within the TAG mound

In a typical mid-ocean ridge setting, the iron content and isotopic composition of the hydrothermal fluid is initially set during high-temperature leaching and alteration of basalt in the reaction zone. This process is suggested to enrich the hydrothermal fluid in iron that is isotopically light (56Fe-depleted) relative to MORB (δ56FeMORB ≈ +0.1‰)23,25,54,55,56. Upon entering and migrating through a large sulfide mound such as TAG, the fluid will be progressively modified through mineral precipitation in open spaces as well as hydrothermal reworking of pre-existing precipitates along its path to the seafloor14,57. However, it remains unclear in how far hydrothermal fluid iron isotope compositions are affected (fractionated) during such interaction24,25,26.

Experimental studies have indicated that chalcopyrite precipitating within seafloor hydrothermal systems rapidly achieves iron isotopic equilibrium with the co-existing fluid and hence chalcopyrite δ56Fe values can be used to assess hydrothermal fluid compositions31. Utilizing the fractionation factor provided by ref. 31 (i.e., Δ56Fechalcopyrite–Fe(aq) = 0.09 ± 0.17‰, 2σ), three vein-related chalcopyrite samples from the lower part of the TAG mound (30–50 m.b.s.f.) yield equilibrium fluid δ56Fe values of −0.13 to −0.01‰ (±0.18‰, 2σ; Supplementary Data 1), which are about 0.1 to 0.25‰ lower than MORB values23,56. The range of our inferred δ56Fefluid values overlaps with and extends to slightly higher values relative to recently (1998) measured vent fluids from the Black Smoker Complex (−0.17 to −0.11 ± 0.02‰, 2σ)28. Since anhydrite within the TAG mound dissolves during inactive periods (i.e., retrograde solubility), the preservation of anhydrite in the chalcopyrite-bearing veins indicates that they formed during the current high-temperature hydrothermal cycle that commenced at ~100 years ago37,39,58,59. Our inferred and the measured28 fluid datasets can thus be interpreted in conjunction and imply that recent TAG hydrothermal fluids experience small negative shifts in their δ56Fe values (<−0.1‰) during ascent through the mound to the black smoker vent site. Such limited evolution of fluid δ56Fe values during upflow, in turn, suggests that the competing effects of rapid pyrite precipitation that lead to increased fluid δ56Fe values versus those of hydrothermal maturation of pyrite and precipitation of chalcopyrite that decrease fluid δ56Fe values (discussed further below) may largely cancel each other out. Alternatively, sub-seafloor sulfide mineralization is volumetrically minor relative to the total flux of dissolved iron through the TAG hydrothermal system. Following venting, oxidation of Fe2+ and precipitation of iron oxyhydroxides within the TAG hydrothermal plume are invoked to cause a more extensive decrease in the δ56Fe values of dissolved iron, down to −1.35‰13,32. This final volume of dissolved iron with low δ56Fe values has been observed to then travel in seawater via currents up to thousands of kilometers away from the TAG site, likely influencing surface planktonic activity in the North Atlantic Ocean13.

Our results now allow us to track the isotopic evolution of dissolved iron throughout the TAG hydrothermal system and reveal an overall and stepwise decrease in δ56Fe values. This finding expands our knowledge on the sequence of (bio-)geochemical processes that contribute to the iron isotopic signature of hydrothermally sourced iron8,24,25,26,27,28,29,32,54 and will thereby help to further refine identification and quantification of such iron in the global oceans and in associated sedimentary records8,13,60.

Evolution of sub-seafloor mineralization at TAG constrained by pyrite δ56Fe values

Based on the δ56Fefluid values from ref. 28 and those calculated in this study, combined with the iron isotope fractionation factors of ref. 30 and ref. 34, the range of δ56Fe values of pyrite in equilibrium with recent TAG hydrothermal fluids is expected to be +0.45 to +1.32‰ (including the 2σ uncertainty; Fig. 3). Notably, only a small subset of our pyrite data plot within this range (6 out of 47; Fig. 3, Supplementary Data 1). However, unlike chalcopyrite, formation of pyrite in seafloor hydrothermal systems has been proposed to occur via transient precursor mineral species that may impose kinetic effects on iron isotope fractionation26,30,34,61,62. On the basis of a synthesis of available experimental and theoretical data, ref. 34 proposed a two-stage model for the formation and iron isotopic equilibration of hydrothermal pyrite at high temperature (>300 °C). In this model, a rapid (few days) initial stage of pyrite formation occurs via an inferred aqueous iron (poly)sulfide precursor phase whose detailed nature (stoichiometry, coordination and magnetic spin) depends on the hydrothermal fluid composition. This inferred short-lived iron (poly)sulfide precursor obtains iron isotopic equilibrium with the fluid, and its δ56Fe signature is then transferred without fractionation upon conversion to pyrite. The initial pyrite will thereby be out of iron isotopic equilibrium with the hydrothermal fluid and will have a δ56Fe value that may be up to ~1.5‰ lower than the value expected for pyrite in equilibrium with the fluid (Δ56Fepyrite–Fe(aq) ≈ 0.8–1‰ at 300–450 °C)30,34. The δ56Fepyrite value will then gradually increase toward this equilibrium value during a subsequent and much slower stage of pyrite recrystallization and iron isotopic equilibration with the hydrothermal fluid30,34.

Fig. 3: Iron isotope results.
figure 3

The δ56Fe values of pyrite (n = 47) and chalcopyrite (n = 3) from TAG-1, TAG-2, TAG-4, and TAG-5, respectively, are plotted as a function of depth below the mound surface. Tie-lines connect sulfide samples obtained from the same drill core piece. Included are the core log for TAG-144 and reference fields for MORB (brown)23,56 and for measured28 and inferred TAG hydrothermal fluids (gray). The inferred range of δ56Fe values of pyrite in equilibrium with recent TAG hydrothermal fluids (pink) was calculated using the fractionation factors of ref. 30 and ref. 34. The lowest δ56Fepyrite value of our dataset (−1.27‰; TAG-4) is used to illustrate a potential initial (disequilibrium) iron isotope composition of pyrite formed via inferred iron (poly)sulfide precursors (dashed black line; cf. ref. 34). We interpret the observed range of δ56Fe values of different textural types of pyrite to reflect contrasting precipitation mechanisms (hydrothermal fluid–seawater mixing vs. conductive cooling) and variable degrees of progressive hydrothermal maturation during the evolution of the TAG complex. See text for details. anh anhydrite, bx breccia, diss. disseminated, py pyrite, sil silica.

Full size image

Rates of ≳1 year to reach pyrite–fluid iron isotopic equilibrium have been estimated under ideal (experimental) conditions34. However, it is not clear what timescales apply to iron isotope equilibration in pyrite in natural SMS systems and how equilibration is influenced by different precipitation mechanisms and by progressive hydrothermal maturation26,52,63. Here we attempt to shed further light on these aspects and integrate our texturally resolved pyrite iron isotope dataset into a refined geologic and hydrothermal framework of TAG15,19,35,36,37,38,39,52.

The lowest δ56Fepyrite values at TAG are found in the massive sulfide mineralization concentrated in the upper parts of the mound (Fig. 3, Table 1). These values are distinctly lower than the estimated δ56Fe values of pyrite in equilibrium with TAG hydrothermal fluids (Fig. 3) and they partly overlap with δ56Fepyrite values reported from seafloor sulfide chimneys elsewhere25,26. The most negative δ56Fepyrite values in our sample suite reach down to −1.27‰ and correspond to porous massive sulfide samples from TAG-4 that exhibit well-developed colloform textures (Fig. 2b). The preservation of such primary depositional features in the massive sulfide and its chimney-like pyrite iron isotope signature are consistent with recent formation at or near the mound surface by growth into open space15. Here, mixing between hydrothermal fluid and cold seawater likely led to rapid precipitation of pyrite with kinetically-driven low δ56Fe values (i.e., strong pyrite–fluid disequilibrium), which have not yet been extensively modified by later hydrothermal maturation (Fig. 3).

Exploring this phenomenon in more detail, we find that pyrite samples from the anhydrite veins in the lower part of the TAG mound (Fig. 2c) have low δ56Fe values that overlap the values of massive sulfide (Fig. 3, Table 1). Notably, the majority of these δ56Fepyrite values are lower than the δ56Fe values of co-existing chalcopyrite from the veins. This confirms iron isotopic disequilibrium in this pyrite group, since pyrite should have higher δ56Fe values than those of chalcopyrite if in isotopic equilibrium (Δ56Fepyrite–chalcopyrite ≈ 0.9‰ at 350 °C)31. The presence of large volumes of associated anhydrite suggests that the pyrite formed recently (≲100 years ago37, see above) due to mixing between hydrothermal fluid and entrained oxygenated seawater within the mound59. Similar to the massive sulfide, the low δ56Fepyrite values of the anhydrite veins can thus be explained by rapid pyrite precipitation followed by only very limited hydrothermal maturation and iron isotopic equilibration with later fluids (Fig. 3).

On the other hand, the coarser-grained pyrite from quartz-pyrite stringer veins in the TAG stockwork (Fig. 2e) have distinctly higher δ56Fe values than those of pyrite from the anhydrite veins (Fig. 3, Table 1). At these deeper levels, i.e. below the anhydrite-rich zone of the mound, entrainment of seawater is decreased and pyrite precipitation is increasingly dominated by conductive cooling of the hydrothermal fluid59. As such, the rates of pyrite precipitation are slower and hence more substantial, albeit not complete, initial pyrite–fluid iron isotopic equilibration could occur during the formation of the quartz-pyrite stringer veins26,34 (Fig. 3). Furthermore, crosscutting relationships confirm that these veins are older than the anhydrite veins53 and have thus likely been subjected to more extensive hydrothermal maturation, leading to further increase in the δ56Fepyrite values30,34.

Pyrite in the sulfide clasts that occur in the different breccia types (Fig. 2d) show δ56Fe values that are overall higher than, but in part overlap the δ56Fepyrite values of the massive sulfide and the anhydrite veins. The values are similar to the δ56Fepyrite values of the quartz-pyrite stringer veins, but are always lower than the estimated δ56Fe values of pyrite in equilibrium with the TAG hydrothermal fluids (Fig. 3, Table 1). Pyrite in these clasts likely have diverse and possibly complex origins that involve combinations of mechanical and hydrothermal reworking of surficial (massive) and vein-related mineralization as well as in situ nucleation and growth of new pyrite15. Such heterogeneous pyrite assemblages should initially have δ56Fe values similar to those of the massive and the vein-related pyrite described above, but the δ56Fepyrite values will progressively shift to higher values as a result of variable degrees of hydrothermal maturation during the protracted development of the TAG breccias, thus offering a sensible explanation for the observed data spread in this particular sample group (Fig. 3).

Remarkably, the finely disseminated pyrite preserved within remnant fragments of altered basalt (Fig. 2e) has the highest δ56Fe values observed at TAG, showing only minor overlap with δ56Fepyrite values of the quartz-pyrite stringer veins and the sulfide breccia clasts (Fig. 3, Table 1). Within individual core samples, disseminated pyrite always has distinctly higher δ56Fe values than those of pyrite that texturally overprints the altered basalt clasts (e.g., stringer veins or massive sulfide cement; Fig. 3). High-temperature hydrothermal alteration of the basaltic host rocks would have commenced during the initial stages of the evolution of TAG and involves chloritization followed by progressive paragonitization and silicification of the basalt, with pyrite forming throughout the alteration sequence38,39. Similar to the quartz-pyrite stringer veins, deeper hydrothermal conditions dominated by conductive cooling lead to slower rates of pyrite precipitation and increased initial pyrite–fluid iron isotopic equilibration. Clasts of variably altered basalt do not only occur in the TAG stockwork, however, but have also been incorporated into the mound breccias at depths much shallower than that expected for the top of the basement. The reason for this is poorly understood, but could potentially be related to high-velocity entrainment in hydrothermal fluid including hydrothermal explosions, or to processes akin to frost jacking and heaving during the repeated expansion (due to internal anhydrite precipitation) and collapse (anhydrite dissolution) of the TAG mound over several high-temperature hydrothermal cycles15,38. Disseminated pyrite preserved in the basalt clasts may thus represent the most extensively reworked sulfide sample material of this study. Such hydrothermal maturation appears to occur intermittently over tens of thousands of years during which the basaltic basement is progressively altered and incorporated into the TAG breccias15,37,38,39, finally imparting a characteristic isotope signature of δ56Fepyrite values in near-equilibrium to equilibrium with the hydrothermal fluids (Fig. 3). Complete iron isotopic equilibration during hydrothermal maturation is potentially aided by the finely disseminated nature of this pyrite, whereas such effects might be restricted to the surfaces of the more massive pyrite types described above.

In summary, we interpret the observed range of δ56Fe values of different textural types of pyrite to reflect contrasting precipitation mechanisms (hydrothermal fluid–seawater mixing vs. conductive cooling) and variable degrees of progressive hydrothermal maturation during the evolution of the TAG mound and stockwork complex (Fig. 3). In contrast to the idealized rates suggested from experiments (≳1 year to reach pyrite–fluid equilibrium34), our results suggest that iron isotopic equilibration of pyrite occurs over timescales of tens of thousands of years within large, dynamic and periodically inactive SMS deposits such as TAG, allowing the preservation of the δ56Fepyrite variations that we observe. The observed iron isotope variations further imply that the W part of the TAG mound (TAG-4) has experienced less extensive hydrothermal maturation than the other parts, consistent with the mineralogical and geochemical asymmetry noted during the original ODP investigation52. Importantly, similar processes can probably explain iron isotope variations in sulfides from fossil onshore VMS deposits (e.g., immature, low-δ56Fe ‘black ores’ and mature, high-δ56Fe ‘yellow ores’), such as the ones found in the classic Kuroko deposits of Japan33. Our study of TAG therefore concludes that sulfide iron isotope compositions can provide insight into the nature, longevity and dynamics of hydrothermal processes in SMS deposits and allow us to create a reference framework for future investigation of similar active and fossil hydrothermal systems elsewhere.

Methods

Sampling

Sub-seafloor samples from the TAG mound and stockwork were sourced from five drill cores (957C, 957E, 957H, 957M, and 957P) originally collected onboard R/V JOIDES Resolution during ODP Leg 158, September–November 199444 (Fig. 1b). Sulfide mineral separates, generally 10 to 100 mg, were extracted from cm-sized drill core pieces using a small-diameter electric drill. Petrographic examination aided the selection of monomineralic sampling sites; in a few cases, however, the fine intergrowths of sulfides only allowed for the recovery of mixtures (pyrite-chalcopyrite or pyrite-marcasite; Supplementary Data 1). The drill tip was repeatedly dipped into ethanol and cleaned using Kimwipes and compressed air between each sample extraction to avoid cross-contamination.

Iron isotope analysis

Iron isotope compositions of sulfide separate (n = 50) were determined at the Vegacenter at the Swedish Museum of Natural History in Stockholm. First, iron-bearing clays and silicates, variably present in separates from altered basalt (n = 8), were removed by dissolution in 10 M HF acid at room temperature for 3 days followed by centrifugation and rinsing of the residue with MilliQ-water. Complete removal of clays and silicates was then confirmed via powder X-ray diffraction (XRD) analysis using the instrumentation and protocol described in ref. 64. The leached and unleached sulfide separates were then dissolved and purified for iron isotope analysis following a procedure adapted from ref. 65 and previously used in ref. 66. About 15 mg of each sample was weighed and transferred to 7 ml Perfluoroalkoxy alkane vials. A volume of 1.5 ml 8 M HNO3 was then added to each sample after which they were evaporated on a hotplate at 70 °C. Once dry, 0.75 ml concentrated HNO3 and 0.5 ml 6 M HCl were added before evaporating the samples again. The evaporated residues were subsequently dissolved in 0.3 M HNO3. For iron separation, aliquots of these solutions (each assumed to contain 200–300 μg Fe) were transferred to clean vials and diluted ten times with de-ionized water to obtain 0.03 M HNO3. Purification was then done by anion exchange chromatography using 100–200 mesh AGMP-1M resin. After iron separation, the samples were dried and then converted to nitric form by repeated dissolution in concentrated HNO3. The samples were finally dissolved in 5 ml 0.3 M HNO3 prior to iron isotope analysis.

Iron isotope analysis was done using an Aridus II Nebulizer system coupled to a Nu Plasma II HR-MC-ICP-MS operated in medium-resolution mode (50 μm slit width, resolving power ~7000). This setup resulted in a sensitivity of ~6–9 V/ppm for δ56Fe for solutions measured at 2 ppm Fe. The sample uptake rate was ~100 μL/min resulting in ~500 μL sample consumption per analysis. The instrumental iron background was 30–40 mV for δ56Fe (~0.2–0.3% of sample intensity) based on on-mass zero measurements of pure 0.3 M HNO3 at the beginning of each run. Each sample was measured six times in a row. The analyses were corrected for mass bias by standard–sample bracketing using the IRMM-014 international standard. Results are reported as δ56Fe and δ57Fe, which correspond to the deviations of 56Fe/54Fe and 57Fe/54Fe relative to IRMM-014 in per mil (Supplementary Data 1). The external reproducibility was 0.07‰ for δ56Fe and 0.10‰ for δ57Fe (2σ), based on repeated measurements of the Alfa Aesar standard solution67 as an unknown throughout the analytical session. Data for samples and standards plot along a mass-dependent fractionation line in a δ56Fe vs. δ57Fe diagram (Supplementary Data 2), confirming that isobaric interferences were properly corrected for. Only δ56FeIRMM-014 values are discussed in the manuscript and figures and all cited literature data have been converted to the same scale.