Evaluating the multiple sulfur isotope signature of Eoarchean rocks from the Isua Supracrustal Belt (Southwest-Greenland) by MC- ICP-MS: Volcanic nutrient sources for early life

On the anoxic Archean Earth, prior to the onset of oxidative weathering, electron acceptors were relatively scarce, perhaps limiting microbial productivity. An important metabolite may have been sulfate produced during the photolysis of volcanogenic SO 2 gas. Multiple sulfur isotope data can be used to track this sulfur source, and indeed this record indicates SO 2 photolysis dating back to at least 3.7 Ga, that is, as far back as proposed evidence of life on Earth. However, measurements of multiple sulfur


| INTRODUC TI ON
Sulfur isotopes have become a valuable tool for reconstructing environmental conditions and biological evolution throughout Earth's history.Sulfur has four stable isotopes ( 32 S, 33 S, 34 S, and 36 S) with abundances of 95.02%, 0.75%, 4.21%, and 0.02%, respectively, which undergo fractionation during numerous biogeochemical processes.So-called mass-dependent fractionation (MDF) occurs during kinetic and equilibrium processes and is controlled by the mass differences between isotopes (Albarède, 2011), and is described by the relationships δ 33 S = 1000•((1 + δ 34 S/1 000) 0.515 − 1) and δ 36 S = 1000•((1 + δ 34 S/1000) 1.91 − 1), where δ 3x S = (( 3x S/ 32 S) sample /( 3x S/ 32 S) VCDT − 1)•1000 and V-CDT is the Vienna-Canyon Diablo Troilite standard.Mass-independent fractionation (MIF) signatures are those that deviate from the typical MDF array.Sulfur isotope compositions are considered to represent MIF when ∆ 33 S (= δ 33 S − 1000•((1 + δ 34 S/1000) 0.515 − 1)) and ∆ 36 S (= δ 36 S − 1000•((1 + δ 34 S/1000) 1.91 − 1)) values diverge from zero (Farquhar & Wing, 2003).MDF and MIF have been particularly valuable for placing constraints on the composition of the Archean atmosphere >2.5 Ga (Johnston, 2011).Farquhar et al. (2000) first recognized that Archean sedimentary rocks preserve a MIF-S signal (expressed in both ∆ 33 S and ∆ 36 S) not seen later in the rock record, signifying a permanent change in the global sulfur cycle.Since these original findings, an abundance of data has been generated verifying this Archean deviation from typical MDF (see e.g., compilation in Claire et al., 2014).The MIF-S signature seen in the Archean sedimentary record is most commonly attributed to the photolysis of SO 2 by ultraviolet radiation and has therefore become known as "smoking-gun evidence" for an anoxic atmosphere lacking an ozone shield (Farquhar et al., 2001;Pavlov & Kasting, 2002;Whitehill et al., 2015).According to this model, volcanogenic sulfur species (SO 2 and H 2 S, initial ∆ 33 S = 0; Siedenberg et al., 2016) are injected into the atmosphere, where they undergo photolysis generating a MIF signal that is carried by the major products, S 8 (∆ 33 S > 0) and H 2 SO 4 (∆ 33 S < 0) (Claire et al., 2014;Ono et al., 2003).Additional products are organic compounds and residual SO 2 that may also be MIF-bearing (see detailed review by Halevy, 2013).Also, selfshielding of the SO 2 molecule during photolysis and polymerization reactions between the products may play a role in controlling the final distribution of isotopic compositions (Endo et al., 2022;Harman et al., 2018).These atmospheric species are rained out and are buried in sediments as sulfide and sulfate minerals, organic sulfur compounds or elemental sulfur.The greatest diversity of products and MIF signatures occurs in non-marine or shallow-marine settings where re-homogenization is minimized (Halevy, 2013).
Detecting MIF-S in ancient sedimentary rocks has therefore long been taken as evidence of atmospheric-derived sulfur during the Archean.A confounding factor in this approach was the discovery of MIF during thermochemical sulfate reduction (Oduro et al., 2011), which occurs at temperatures exceeding 120°C (reviewed by Cai et al., 2022).The relevance of this process to ancient sedimentary data needs to be assessed on a case-by-case basis.
Despite its complexity, the sulfur isotope proxy remains one of the key tools for reconstructing the ancient sulfur cycle and, therefore, also carries great importance for understanding the origin and early evolution of life.Microbial sulfate reduction is an ancient energy-yielding metabolism (Ueno et al., 2008), but prior to the Paleoproterozoic Great Oxidation Event (GOE, ca.2.3 Ga), sulfate concentrations were likely several orders of magnitude smaller than today (Crowe et al., 2014) and volcanism constituted the major source of sulfate to the biosphere (Muller et al., 2016).One may therefore speculate that a thriving early biosphere would have been enabled by volcanic sulfur sources.
The oldest proposed evidence of life on Earth goes back to ca. 3.7-3.8Ga and is preserved in meta-turbidites and carbonate rocks in the Isua Greenstone Belt (e.g., Mojzsis et al., 1996;Nutman et al., 2016;Ohtomo et al., 2014;Rosing, 1999).Sulfur isotope analyses of banded iron formations and other schists of the Isua Greenstone Belt have indeed revealed the presence of MIF (Papineau & Mojzsis, 2006;Whitehouse et al., 2005), suggesting an atmospheric source of sulfur and supporting their interpretation as meta-sedimentary rocks.However, no sulfur isotope data so far exist from those rocks that hold the most promising biosignatures.The biogenicity of the carbonate record in particular is debated (Allwood et al., 2018;Zawaski et al., 2020), leaving the possibility that the carbonates are of abiotic hydrothermal origin.Additional multiple sulfur isotope analyses on these low sulfur content biosignature-bearing rocks would help distinguish if the major sulfur-bearing phases were derived from the atmosphere, supporting a sedimentary setting.Furthermore, if there is a signature of atmospheric sulfur input, and if life existed at that time, microbial sulfate reduction could have been a viable metabolism.To further validate this point, additional sulfur isotope analyses are required using small sample size methods.
Traditionally, sulfur isotopes are measured using gas-source isotope-ratio mass spectrometry (GS-IRMS), for which samples are converted to SO 2 or SF 6 gas (Hulston & Thode, 1965).While some initial studies using this method reported sample size requirements of 100 nmol to 5 μmol sulfur (Ono et al., 2006;Paris, Fehrenbacher, et al., 2014) more recent workers have been able to push this limit down to 20 nmol (Au Yang et al., 2016).Alternatively, an ion microprobe can be used for measuring sulfur isotopes of geological samples in situ (Mojzsis et al., 2003;Whitehouse et al., 2005), and this technique has even been optimized for finely disseminated sulfides in S-poor rock samples (Bryant et al., 2019;Marin-Carbonne et al., 2022).These in situ techniques are vital for understanding multiple generations of sulfides that can exist in Archean rocks, and for distinguishing fluid-rock interactions and diagenetic and recrystallization histories (Mojzsis et al., 2003;Whitehouse et al., 2005).Where these methodologies are not available, multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) has been developed for triple sulfur isotope ( 32 S, 33 S, 34 S) measurements with sample size requirements down to 10 nmol of sulfur at a precision of 0.10‰ for ∆ 33 S (Burke et al., 2018;Paris et al., 2013;Rennie et al., 2018).This technique has been successfully applied to recent environmental samples  (Burke et al., 2018, Rennie et al., 2018), as well as to Neoarchean carbonate-associated sulfate (CAS) (Paris, Adkins, et al., 2014).an aqua regia leaching protocol that can be applied to shales and volcanic rocks.We applied this method to those meta-turbidite and carbonate rocks that have been put forward as key archives of early life (Nutman et al., 2016;Rosing, 1999).To further validate this method, we also analyzed a range of standards, as well as two samples of Proterozoic shale from the Torridon Group in Scotland (1.0 Ga), where more material was available and no MIF should be present.To support our interpretation of the Eoarchean sulfur isotope data, we performed detailed optical microscopy, XRF mapping, bulk elemental analyses, and geochronology of selected mineral phases, which allowed us to evaluate primary and secondary controls on the samples.Although alteration is impossible to rule out, the most parsimonious explanation of our data is a primary atmospheric source of sulfur in these rocks at 3.7 Ga, confirming the idea that organisms living at that time would indeed have had access to this nutrient source.

| Regional geology of the Isua Supcracrustal Belt
The Isua Supracrustal Belt (ISB) is located within the Itsaq Gneiss Complex (IGC) in the southwest of Greenland (Figure 1).The complex is mainly composed of gneisses derived from tonalite-trondhjemitegranodiorite (TTG), granitic, and dioritic protoliths which have undergone multiple deformation events and high-grade metamorphism (Nutman et al., 1993).However, around 10% of the complex comprises metasedimentary, mafic, and ultramafic rocks (Nutman et al., 1996).The best-preserved examples of these varied rock types lie in the north of the complex which includes the ISB (Nutman et al., 2002).The ISB is the largest supracrustal inclusion within the IGC at 35 km long and up to 4 km wide (McGregor & Mason, 1977).
Using U-Pb dating of zircon, Nutman and Friend (2009) have differentiated the ISB into two terranes: a southern ~3800 Ma terrane and a northern ~3700 Ma terrane.Both terranes contain a diverse range of rock types such as felsic schists, amphibolites of differing compositions, and banded iron formations (Nutman & Friend, 2009).
Parts of the ISB have experienced only up to amphibolite facies metamorphism and contain rare areas of low strain (<5% of the belt), so these preserve features such as pillow lavas and sedimentary structures unlike other global Eoarchean terranes (Nutman et al., 2022).
The presence of pillow structures and BIFs suggests deposition in a submarine setting (Appel et al., 1998).Early work revealed that the ISB underwent regional metamorphism before 3600 Ma at prograde temperatures and pressures of at least ~550°C and ~5 kbar, buried to at least 15 km depth (Boak & Dymek, 1982).Ramírez-Salazar et al. (2021) later identified three metamorphic events, including an amphibolite facies event in the Eoarchean followed by a lower amphibolite facies event in the Neoarchean and finally a later lower grade retrogression.Zuo et al. (2021) contribute further deformation evidence, possibly linked to heat-pipe tectonics during emplacement.Despite this complex metamorphic history, Isua remains an essential locality for understanding environments and potential life on the early Earth.

| ISB meta-turbidites
The meta-turbidite samples analyzed in this study (sample numbers 208288, 208290, 208293, and 208294) were collected during a field trip in 2019 from a felsic schist unit on the western side of the ISB in the ~3700 Ma terrane (Figure 1).The locality for these samples was described by Rosing (1999) as ~50 m across with predominantly 10-70 cm graded beds mixed with 10 cm slate units.The package has been interpreted as a turbidite sequence.Most notably, rocks from this locality contain graphite, which has been found to have a depleted δ 13 C signature of −19‰.This signature could potentially represent some of the earliest evidence of microbial life on Earth (Rosing, 1999).Furthermore, one sample contains a molybdenite vein, which is intriguing because molybdenum is an essential ingredient for the evolution of life and it has been suggested to be key in the coevolution of life and environment (Anbar & Knoll, 2002;Schoepp-Cothenet et al., 2012).The specific samples targeted in this study are slates which have been previously analyzed for major and minor elements along with carbon and nitrogen isotopes and total nitrogen, sulfur, and organic carbon content (Stüeken et al., 2021).
They are two separate off-cuts of a ca.20 cm large block.Nutman et al. (2016) proposed that this outcrop locality contains primary meta-carbonate containing fossil stromatolites, which would be the earliest physical evidence of life on Earth, predating the next-oldest example of stromatolites by 200 Myr (Wacey, 2010).Nutman et al. (2016) suggest that the presence of a CO 2 -rich fluid during metamorphism prevented reactions between quartz and dolomite, allowing sedimentary structures to remain intact.This mechanism, along with additional evidence such as the presence of internal laminae, trace element compositions, and adjacent shallow water storm deposits were used to argue in favour of a biogenic origin.However, these claims have been disputed by Allwood et al. (2018) using finer-scale geochemical and threedimensional analysis.They documented characteristic trace element compositions not only in carbonate layers but also in silicate layers and therefore attributed the carbonate in this rock to metasomatic alteration rather than a primary sedimentary assemblage.This was further supported by detailed structural analysis by Zawaski et al. (2020), who concluded that the structures are boudins formed through tectonic activity rather than stromatolites.
Many ISB rocks were originally interpreted as primary sedimentary carbonate but have now been linked to metasomatism (Rose et al., 1996).The use of sulfur isotopes may help to distinguish a sedimentary origin for this locality as Archean sediments should preserve a MIF-S signature.Nutman and Friend (2009).(d) Geological map of sample area on the eastern side of the ISB.Modified from Nutman and Friend (2009).

| ISB amphibolites
We also obtained two samples of amphibolite from the ISB to serve as control samples of non-sedimentary origin.Sample 208399 is collected from the same site as the meta-carbonate described in Section 2.3, while 207849C is from an adjacent amphibolite unit.These samples have not been previously studied in the literature, so no existing data are available.As these rocks are derived from a mafic igneous protolith, no MIF-S signature is expected and should reflect the mantle signature of ∆ 33 S = 0 (Siedenberg et al., 2016).As a result, these samples are used to compare meta-carbonates and meta-turbidites thought to be of sedimentary origin.Their proximity to the meta-carbonates may also be useful in identifying potential later metasomatism.

| Proterozoic control samples
To test the reproducibility of the MC-ICP-MS sulfur isotope method on whole-rock samples, the Eoarchean samples from the ISB were supplemented with two samples of late Proterozoic age (samples 180808-1 and 180808-14) from the Diabaig Formation of the Torridon Group in the northwest of Scotland.Like the ISB rocks, these samples have a complex matrix and relatively low sulfur content, making them suitable for testing the new sample preparation protocol.The samples have been previously measured for δ 34 S (Nielson et al., 2024) using EA-IRMS.They postdate the Archean and, therefore, should show no MIF-S.As a result, they could be used to verify that the method is accurate for whole-rock samples with a complex matrix.These samples were prepared twice each to test the reproducibility of the method.

| Cutting and pulverizing
All sample preparation for sulfur isotope analyses was carried out at the School of Earth and Environmental Sciences at the University of St Andrews.The ISB meta-turbidites and the Proterozoic shales had already been powdered for a previous study (Nielson et al., 2024;Stüeken et al., 2021).Hand samples of the ISB meta-carbonates and amphibolites were cut to an appropriate size and weathered surfaces were removed using a water-cooled diamond saw.The blocks were hammered into sub-cm-sized pieces using a steel mortar and pestle.These chips were milled into a fine powder with a Fritsch Pulverisette agate ball mill.Between samples, all equipment was cleaned with deionized water and ethanol and dried with compressed air.Additionally, between samples the ball mill was run for several minutes with pre-combusted silica sand to clean out any excess residue.

| Thin section petrography
Off-cuts of all Archean samples were cut to size for thin sections and polished with grinding laps.General petrographic analysis of mineralogy and textures, including reflected light microscopy, was carried out using a Leica DM750 P polarization microscope at the University of St Andrews.Photomicrographs were taken using a Leica ICC50 W microscope camera module and processed using Leica LAS EZ imaging software.

| Bulk rock sulfur content
To determine the approximate abundance of sulfur in each sample, approximately 20 mg of sample powder, plus 5 mg of V 2 O 5 powder, were weighed into 8 mm × 5 mm tin capsules (Thermo Fisher) and analyzed by flash combustion with an EA Isolink (Thermo Fisher) coupled to a MAT253 isotope-ratio mass spectrometer via a Conflo IV (Thermo Finnigan).In the EA, the combustion furnace mantling the reactor was set to 1020°C.Pure O 2 gas was injected at a flow rate of 250 mL min −1 for a duration of 5 s starting with the drop of the sample.The reactor was filled with 6 cm anhydrous tungstic oxide granules as an additional combustion aid, followed by 13 cm of Cu wire as a reductant of excess O 2 .The gas stream was passed through a water trap filled with Mg(ClO 4 ) 2 at room temperature before entering the gas chromatography column (GC).All reagents were purchased from Elemental Microanalysis.The temperature of the GC was ramped from 60°C to 240°C to ensure complete elution of all gases.The measured SO 2 peaks were too small (<5 vs.) to obtain accurate isotopic analyses, but approximate sulfur abundances could be estimated.Abundances were calibrated against peak areas generated by pure sulfide standards IAEA-S2 and IAEA-S3.This method was also used to obtain sulfur isotope analyses by IRMS of the standard NBS-127.This standard was analyzed by MC-ICP-MS as well, providing additional cross-calibration between the two techniques.

| Sulfur leaching and purification for isotopic analyses
Wet chemistry and isotopic analyses were carried out in the St Andrews Isotope Geochemistry (STAiG) clean laboratories.Sulfur was leached from the samples with aqua regia as described in Xu et al. (2012).Approximately 100 mg of each whole-rock sample powder was weighed into acid-cleaned Teflon cups.To each cup, 3 mL of concentrated HCl and 1 mL of concentrated HNO 3 were added.
The cups were then capped and left on a hotplate overnight at 80°C.The contents of each cup were then transferred into a 15 mL acid-cleaned Falcon tube.The samples were then centrifuged at 3900 rpm for 10 min to separate the leachate from the residue.The leachate was then transferred into clean Teflon cups.The cups were left uncapped on a hotplate overnight at 80°C to evaporate.The resulting dried solids were re-dissolved in 2 mL of 0.5 M HNO 3 and left on a hotplate at 80°C for an hour to ensure that samples were completely dissolved.The samples were then transferred into 2 mL disposable centrifuge tubes and centrifuged at 1300 rpm for 5 min to separate any remaining sediment.Using a pipette, 1.8 mL of each sample was extracted for subsequent use.Guided by approximate sulfur abundances measured by EA-IRMS (Section.3.3), dilutions of each sample were prepared and analyzed for sulfur content using an Agilent 8900 ICP-QQQ, run with the collision cell in O 2 mode (Nakano, 2018).
An aliquot of solution containing 20-100 nmol of sulfur was pipetted into Teflon vials.These aliquots were placed on a hot plate at 104°C for a few hours until just evaporated.Residues were then re-dissolved in 4 mL 0.01% HCl and transferred into 15 mL centrifuge tubes.These solutions were then purified by anion exchange chromatography (resin AG1-X8) in a PrepFAST MC unit fitted with a 2 mL column, and a flow rate set to 2 mL min −1 .The resin was washed and conditioned for each subsequent sample by washing with 20 mL 1.1 M HCl, 20 mL MilliQ water, 20 mL1.6 M HNO 3 , a further 20 mL 1.1 M HCl and finally conditioning with 20 mL 0.06 M HCl before sample loading.The sample matrix was washed with 30 mL MilliQ water before the sulfate fraction was collected in 12 mL of 0.5 M HNO 3 .These purified solutions were dried down at 104°C, before being re-dissolved in 80 μM NaOH (250 μL of 2 μM NaOH for blanks).At the end of the sequence, the column is washed with 20 mL 1.1 M HCl and 20 mL MilliQ water before being stored in 0.1 M HCl.

| Sulfur isotope mass spectrometry
Isotope measurements were carried out with a Neptune Plus MC-ICP-MS, using sample-standard bracketing to correct for instrumental mass bias and drift.Samples were concentration-checked prior to isotope analysis and further diluted with 80 μM NaOH (the final volume of which depended on the actual number of nmol of SO 4 in the sample based on the concentration check) to achieve a matrix match with an in-house 40 μM Na 2 SO 4 bracketing standard (Paris et al., 2013).Before each run, the instrument was tuned to optimize signal sensitivity and mass resolution and reduce interferences from 32 S-1 H to 16 O-16 O.This was done by adjusting and optimizing the torch position, source lenses, gas flows, and focus quad (Paris et al., 2013).Samples were run over several sessions and all samples were measured in triplicate where possible (and at least twice) to give an estimate of analytical reproducibility.Measurements are reported using delta notation calculated as follows relative to Vienna-Canyon Diablo Troilite (V-CDT), where x = 33 or 34: ∆ 33 S values were calculated using the equation: Samples are considered to display MIF if ∆ 33 S values fall outside the range from −0.14‰ to +0.14‰, which is based on the longterm external reproducibility of ∆ 33 S on full procedural replicates of standards using this method.Full procedural blanks (including dissolution) along with blanks from column chromatography were run alongside samples to track any potential contamination.Standard values were blank-corrected using column chromatography blanks, while sample data were corrected using full procedural blanks generated in this study (Table S1).Accuracy was evaluated with the external standards NBS-127 (barium sulfate), IAEA-SO-6 (barium sulfate), and IAEA-SO-5 (barium sulfate), as well as three in-house standards: Switzer Falls river water (Burke et al., 2018(Burke et al., , 2019)), a crushed and homogenized deep-sea coral and seawater from the North Sea-see Section 4.2.1 for the results of these tests.

| Micro-XRF element maps
To further characterize the samples, in particular, the distribution of major and minor elements that may aid in mineral identification and tracking secondary alteration, separate slabs of three metaturbidites (208288,208290,208294), the meta-carbonate block, and one amphibolite sample (208399) was analyzed by micro-XRF.
A Bruker M4 Tornado Plus micro-XRF instrument (Bruker Nano GmbH, Berlin) was used at the University of Copenhagen.The instrument is a non-destructive analyser that enables fast determination of elements and their distribution on a flat sample surface (Flude et al., 2017).The micro-XRF instrument uses a Rh X-ray source with a polycapillary lens that focuses the X-rays in combination with a collimator for reducing the aperture.The signal is captured with two XFlash® silicon drift detector (130 eV resolution).The beam size can go down to 20 μm and elements from Na to U can be detected (Bruker, 2023).The micro-XRF instrument detects individual elements by fluorescence X-rays (secondary X-rays) emitted when the sample is bombarded with high-energy X-rays.From Moseley's law, the peak of the lines increases approximately at the square root of the atomic number from the emitted element, so, for example, Fe has a higher intensity than Si (Wenk & Bulakh, 2016).The maximum depth of recorded fluorescence X-rays, the XRF saturation depth, varies with atomic number (Janssens et al., 2000).Minerals with heavier elements thus have a deeper XRF saturation depth than minerals with lighter elements.The XRF signals attenuate by depth and the XRF from Si has lower energy and is attenuated by the surroundings (Flude et al., 2017).
The measured samples were placed in the sample chamber under low-vacuum conditions (2 mbar) to avoid Ar absorption and to facilitate the detection of light elements.The Rh X-ray tube energy was set at 50 kV and a current of 600 μA.Measurements were performed using a 20 μm step size, 20 μm beam size, and an acquisition time of 20 ms per pixel.Spectral quantification was performed in standardless mode using M4 Tornado software.The resulting element

| In situ molybdenite Re-Os dating
Element mapping revealed the presence of a molybdenite-filled vein in one of the meta-turbidite samples (Figure S1).This would be a significant host of sulfur in the rock and can be dated with Re-Os geochronology, allowing to constrain the timing of a major phase of sulfur introduction into the turbidites.In situ Re-Os isotopic analysis was conducted at Adelaide Microscopy (The University of Adelaide) using a RESOlution-LR 193 nm excimer laser ablation system (Applied Spectra), with a S155 sample chamber (Laurin Technic), coupled to an Agilent 8900x ICP-MS/MS.Analytical conditions were identical to Tamblyn et al. (2024).A gas mixture of CH 4 in He was used as the reaction gas, optimized for 187+14 Os reaction rates, while suppressing interference of 187+14 Re (Hogmalm et al., 2019;Tamblyn et al., 2024).N 2 gas (3.5 mL min −1 ) was added to the carrier gas after the sample chamber to increase sensitivity (Hu et al., 2008).The following isotopes were measured, with dwell times in milliseconds between brackets: 34 S(2), 95 Mo(2), 185 Re(20), 185+14 Re(50), 187 Os(50), 187+14 Os(100), 189 Os(50), 189+14 Os(100). 185Re was measured as a proxy and 187 Re calculated assuming natural isotopic abundance.The laser fluence was set at 3.5 J cm −2 with a repetition rate of 5 Hz and a spot-size of 100 μm (except for the high-Re primary standard where a spot-size of 43 μm was used).Analyses included 30 s of background collection followed by 40 s of ablation.Data reduction was conducted in the LADR software (Norris & Danyushevsky, 2018) using molybdenite MDQ0252 from the Merlin deposit as the primary reference material for the Re/Os ratio calculations with an ID-TIMS 187 Os/ 187 Re ratio of 0.025649 ± 0.000105 (Tamblyn et al., 2024).

| Mineralogy and thin section petrography
For a summary of overall mineral components for all ISB samples as measured by XRD, see Table S2.Textural relationships are described in the following:

| ISB meta-turbidites
Samples 208288 and 208294 are both finely crystalline and composed predominantly of quartz and mica.The rocks display strong foliation defined by biotite and chlorite mica.However, chlorite is seen to replace biotite crystals indicating it formed at a later stage (Figure 2a).Sample 208288 is permeated by secondary veins of coarser quartz crystals (Figure 2b).Both samples display speckled occurrences of graphite (Figure 2c).This could be a potential source of organic-bound sulfur in the rock.Furthermore, sulfide minerals, most likely chalcopyrite, are visible under reflected light in both samples (Figure 2d).
Sample 208290, similarly to the previous sample is highly foliated and composed mainly of quartz, biotite, and chlorite.Foliated layers in this sample, however, appear more intensely deformed with several small-scale fold structures indicating a high amount of strain and deformation.This sample contains thin secondary quartz veins along with large patches of carbonate (Figure 2e).Foliation both wraps around and is disrupted by the carbonate minerals, suggested that the carbonate formed during secondary alteration or metasomatism.Graphite is again present throughout the sample as a potential sulfur source.The most common opaque minerals (Figure 2f) may be magnetite or ilmenite grains.

| ISB amphibolites
Amphibolite sample 208399 is mostly composed of amphibole, quartz and plagioclase feldspar with minor accessory carbonate and opaques.The sample contains large amphibole crystals >500 μm across, some of which are split into sub-grains characteristic of deformation (Figure 4a).Accessory carbonate is likely a sign of secondary alteration.Under reflected light, some of the opaque minerals can be identified as chalcopyrite (Figure 4b).Sample 207849C has a similar composition to 208399 with less amphibole and a greater abundance of quartz.An interesting feature of this sample is a large vein (ca. 0.4 mm width and at least several cm length) cross-cutting the entire section.Most of the vein is filled with zoisite (Figure 4c).
Also, smaller vein structures (0.2 mm width and a few cm long) are filled in with zoisite.The large vein further contains sulfides, likely chalcopyrite and pyrite (Figure 4d).Additional sulfides are present outside of the vein.The presence of these veins and secondary sulfides along with carbonate point to later fluid infiltration into the rock.

| Standards and blanks
Across the range of external and in-house consistency standards with δ 34 S values ranging from −34‰ to 22‰, our method agrees well with previous literature (Böttcher et al., 2007;Brand et al., 5, Table 1).For each standard, the average of the full procedural replicates is within two standard deviations of the previously reported values.External reproducibility based on standards that had more than three replicates for δ 34 S is 0.12‰ (2 SD) and for ∆ 33 S is 0.08‰, which is comparable or better than previous studies.Full procedural blanks range from 0.30 to 0.73 nmol of sulfur with an average of 0.37 ± 0.30 nmol (2 SD; n = 8) and had an average δ 34 S = 9.20 ± 3.49‰ (2 SD; n = 8) while the blank of the prepFAST chromatographic column alone contained 0.22 ± 0.14 nmol of sulfur (2 SD; n = 11) and had an average δ 34 S = 4.60 ± 3.51‰ (2 SD; n = 11) (Table S1).

| Proterozoic shales
For the two Proterozoic shales that were used for method validation (Table S3), full procedural replicates fall within the error range of each other and previously measured δ 34 S values; however, we note that errors on existing measurements, which were obtained by EA-IRMS, are relatively large due to the low S abundance in these rocks.
This indicates overall good reproducibility of the method including the digestion steps.

| Elemental composition and mapping
Bulk elemental compositions are shown in Table 3 and Table S4.We calculated Eu anomalies as proxies for hydrothermal input, following the procedure of Schier et al. (2020): Here SN = normalized to Post-Archean Australian Shale from McLennan (1989).The meta-turbidite samples show (Eu/Eu*) SN of 0.869-1.970,that is, a high degree of variability (Figure 6a).In contrast, the meta-carbonate and amphibolite samples all exhibit positive Eu anomalies ranging from 1.376 to 2.980 (Figure 6b), similar to other Archean chemical sediments (Viehmann et al., 2015) and consistent with previous reports from this carbonate unit (Nutman et al., 2016).Ratios of iron to aluminum, a proxy for redox conditions, and hydrothermal input, show a mean value of 0.63 ± 0.17 for the meta-turbidites (Figure 6c), consistent with deposition under anoxic conditions (Raiswell et al., 2018;Stüeken et al., 2021).As pointed out by Raiswell et al. (2018), hydrothermal Fe enrichments are typically manifested in Fe/Al ratios exceeding 2.0, which is not the case in our samples.We therefore tentatively propose that these values were not strongly affected by hydrothermal alteration.For the metacarbonate, four measurements cluster around a mean of 3.94 ± 1.48 with one outlier at 191.43.The amphibolite shows a value of 3.75, which is higher than the Fe/Al ratio of modern oceanic crust (1.03;  (Ptáček et al., 2020), and found the amphibolite as well as the metacarbonate to cluster close to the composition of modern oceanic crust (Figure 6d).This contrasts with the composition of the metaturbidites, which appear to include relatively greater proportions of both ultramafic and felsic crust.We stress that secondary processes can alter these ratios.This interpretation is therefore tentative.
Another noteworthy geochemical feature is Ba concentration above the limit of the detector (>1 wt.%) in two meta-carbonate analyses, likely indicating the presence of Ba-carbonate or possibly Ba-rich mica; as noted above, no barite was detected by XRD or microscopy.
Element maps of the meta-turbidites (Figure S1a,b in Appendix S3) are consistent with graded lamination and variable mixing between clays and sand particles.Sample 208290 shows lamina-parallel enrichments in Ca (Figure S1a), probably due to the presence of (possibly secondary) carbonate.Sample 208294 contains a lamina-parallel enrichment in Mo (Figure S1b), likely reflecting a secondary injection F I G U R E 5 (a) Average δ 34 S values obtained for standards in this study vs. literature values (Böttcher et al., 2007;Burke et al., 2019;Paris et al., 2013).(b) δ 34 S anomalies of individual replicates measured relative to average standard values obtained in this study.

TA B L E 1
Sulfur isotope data for external and in-house standards.
of molybdenite.Some of the molybdenite from this rock was used for in situ Re-Os dating.Among the amphibolites, sample 208399 shows a clotted texture (Figure S2a), likely reflecting metamorphic recrystallization of original magmatic minerals.Stringers of Si enrichment are probably the result of minor secondary quartz veins.
Sample 207849, which displays greater proportions of quartz/zoisite veining at hand sample scale, also shows strongly Si-enriched bands in elemental maps (Figure S2b).Lastly, the elemental map of the meta-carbonate sample reveals banding of quartz and carbonate (Ca, Fe, Mg) (Figure S2c).Strongly Al-enriched laminae may reflect stylolites.

| In situ Re-Os geochronology results
All data obtained from the Re-Os analyses are provided in Appendix S1.For the reference materials, common Os was below detection limits and, therefore, Re-Os dates were directly calculated as weighted mean ages from the 187 Re/ 187+14 Os ratios.The excluding the three younger outliers, anchored to a present-day initial 188 Os/ 187 Os ratio set at 6.74 ± 1.00, produced a Re-Os date of 3662 ± 214 Ma (Figure 7d).

| In situ Rb-Sr geochronology results
Detailed results from the Rb-Sr analyses are presented in Appendix S2.
MDC phlogopite was used to correct the Rb/Sr ratio for matrixdependant fractionation, using the method described in Glorie, Gilbert, et al. (2024).The reference age for MDC is 519.4 ± 6.5 Ma and the inverse isochron was anchored to an initial 86 Sr/ 87 Sr ratio of 1.3773 ± 0.0013, constrained from a diopside (low-Rb mineral) that occurs in the same location (Hogmalm et al., 2017) (Figure 8a).The offset in the measured isochron lower intercept Rb/Sr ratio compared to the reference value, calculated from the Rb/Sr age for MDC, was used to correct the Rb/Sr ratios for all other analyzed samples.For the Bund-1b and Taratap secondary reference materials (Figure 8b,c), biotite, K-feldspar, and plagioclase (± apatite) were analyzed for Rb/ Sr and Sr/Sr ratios, yielding matrix-dependant fractionation corrected inverse isochron dates of 286.8 ± 2.1 and 499.5 ± 3.6 Ma, respectively.

| Evaluating the effectiveness of the MC-ICP-MS method
All standards produced accurate values in agreement with previous measurements in the literature, with reproducibility comparable or better than previous studies (Table 1), illustrating the accuracy and   S3), and they did not display MIF, as expected for the Mesoproterozoic.Of note is a slight bias to positive ∆ 33 S in the standards and replicate Proterozoic samples, with the average value across all standards for ∆ 33 S = 0.06 ± 0.08‰ (2 SD; n = 20) and the average value across the Proterozoic shale samples for ∆ 33 S = 0.07 ± 0.10‰ (2 SD; n = 4).Although we do not yet know the reason for this slight positive bias, we note that ∆ 33 S is still within the error of zero and is much smaller than the signals measured in this study, so it does not impact on our conclusions.In consistency with these results, we interpret anything within 0.14‰ of 0‰ as representing mass dependent fractionation.Future work will investigate potential causes for this slight positive ∆ 33 S.
For replicate analyses of the same ISB solution on the MC-ICP-MS, we obtained average errors of 0.14‰ for δ 34 S (2σ) and 0.23‰ for ∆ 33 S (2σ), which is slightly improved over ion microprobe studies (Muller et al., 2016;Papineau & Mojzsis, 2006).Collectively, these results give confidence in the accuracy and precision of our analytical method.Procedural blanks prepared alongside the ISB produced an average of 0.37 ± 0.30 nmol (2 SD; n = 8) of sulfur with the largest blank being 0.73 nmol.The prepFAST column chemistry blanks have an average sulfur content of 0.22 ± 0.14 nmol (2 SD; n = 11) (Table S1).Overall, our complete procedural blanks were thus only slightly larger than prepFAST column chemistry alone, but comparable with existing studies using column chemistry and sulfur isotope measurement by MC-ICP-MS with values ranging from 0.16-0.60 nmol and an average of 0.38 ± 0.38 (2 SD; n = 5, Paris, Adkins, et al., 2014).Given our typical sample size was 50 nmol, this blank represents only ~0.74% of the sample, therefore, the blanks are not a concern for the whole-rock digestion method, making it applicable to samples with very low sulfur contents (a few tens of μg/g in our case).Since aqua regia is an oxidizing agent, it is likely to dissolve organic-bound sulfur and sulfide minerals, and those were probably the main sulfur phases in our samples.Aqua regia does not fully dissolve barite (Abbasi et al., 2016); however, barite was not visible in our samples and not detected by XRD (Table S2, Section 4.1.2),and so this issue would likely not be relevant for this study.Additional method development would be required for rock samples containing significant barite.The data presented here are most likely dominated by sulfide minerals and possibly organic-bound sulfur.

| Comparisons to existing Archean sulfur isotope data
Our ISB meta-turbidite data (both ∆ 33 S and δ 34 S) align well with previously published data from banded iron formations (BIFs) and garnetbiotite schists in the ISB reported by Papineau and Mojzsis (2006) (Figure 9a).Their measurements were made using an ion microprobe targeting the individual sulfide minerals cubanite, chalcopyrite, pyrite, and pyrrhotite.Those sulfides were largely taken from metasedimentary units in the north-eastern segment of the ISB with one sample locality matching the western felsic schist unit examined in this study.
This similarity suggests that the sulfur in the meta-turbidites, BIFs, and garnet-biotite schists in the ISB may have come from the same source.The presence of MIF indicates an atmospheric contribution that was generated in an oxygen-poor atmosphere and it would imply that volcanism delivered S-metabolites to life, if present.More specifically, the positive ∆ 33 S values suggest that the atmospheric carrier of the MIF signal was elemental sulfur (S 8 ) that eventually became incorporated into sediments as sulfide (Ono et al., 2003).A sedimentary origin for these Eoarchean rocks is therefore plausible.In terms of δ 34 S, the meta-turbidites show a very limited spread, consistent with Eoarchean δ 34 S literature values ranging from ~−7.5‰ to 8‰ (Morrison & Mojzsis, 2020 and references therein).This narrow spread contrasts with the Neoarchean (range from ~−25‰ to 37‰) and may reflect a relatively small sulfur reservoir in seawater with limited scope for biological isotope fractionation (Morrison & Mojzsis, 2020).This interpretation is consistent with the inferred anoxic conditions based on Fe/Al ratios (Figure 6c).However, the sulfur isotope data from these rocks alone cannot rule out post-depositional introduction of younger Archean sulfur with a MIF signal.We will further address this issue below (Section 5.3).
The meta-carbonate rocks exhibit a negative ∆ 33 S signature, unlike the ISB meta-turbidites, schists and BIFs (Figure 9a,b).Similar negative ∆ 33 S values have been documented from Archean barite deposits at 3.25-3.55Ga (Bao et al., 2007;Farquhar et al., 2000;Roerdink et al., 2012;Ueno et al., 2008) (Figure 9a).As stated in Section 4.1.2,no barite was identified in the thin section or by XRD.Furthermore, since our aqua regia leaching protocol is not optimized for barite, and since sulfide minerals were visible in the samples, the negative ∆ 33 S values are most likely sulfide-hosted.
In this case, they may reflect a distinct source of sulfur (temporal or spatial) compared to the meta-turbidites, schists, and BIFs.
The most plausible explanation for these values is reduction of seawater sulfate.Transfer of negative ∆ 33 S from sulfate to sulfide has been well documented in slightly younger Paleoarchean successions at 3.5-3.2Ga, where it is usually attributed to microbial sulfate reduction (e.g, Roerdink et al., 2013;Ueno et al., 2008).
In fact, this mechanism may explain many occurrences of nega-  Whitehouse, 2007;Kaufman et al., 2007;Mishima et al., 2017;Ono et al., 2003;reviewed by Claire et al., 2014;Johnston, 2011).We cannot infer the presence of this metabolism from our data alone (∆ 36 S measurements or a larger sulfate reservoir enabling larger MDF would be needed), but it is a plausible explanation and would confirm the volcanic supply of a key metabolite for the early biosphere at that time.The presence of MIF in the meta-carbonates may reflect a sedimentary origin for these rocks, consistent with the REE data (see Allwood et al., 2018;Nutman et al., 2016, for further debate on this topic), but as with the meta-turbidites, we cannot easily rule out post-depositional introduction of younger Archean sulfur with a MIF signature (Section 5.3).

| Evidence and effects of metamorphism and alteration
Processes undergone by the ISB rocks such as subsequent metamorphism and alteration have the potential to modify original sulfur isotope signatures and, therefore, must be carefully considered.Increasing metamorphic grade has been shown to impart a positive fractionation on δ 34 S of up to 2-3‰ at high-grade conditions (amphibolite or above; Bucholz et al., 2020).Furthermore, regional metamorphism may be capable of homogenizing sulfur pools and thus reduce the ∆ 33 S MIF signature seen in Archean meta-sedimentary sulfide minerals (Cui et al., 2018).The abundance of amphibole and the absence of high-grade mineral associations such as sillimanite and K-feldspar confirm that the samples experienced up to amphibolite facies conditions, and hence some fractionation in δ 34 S due to metamorphism may have occurred, but since the metamorphic fractionation 2-3‰ is overall smaller than microbial sulfate reduction (up to 70‰; Sim et al., 2011), it does not impact our interpretation.
Of greater concern is secondary alteration of the MIF-S sulfur isotope signature by fluids.Confirmation of alteration of the ISB rocks is indicated by the presence of replacement chlorite, secondary carbonate and vein structures filled by quartz, sulfides, and zoisite (Figure 4, Figure S1).The replacement of biotite by chlorite is a common reaction in many rocks and is observed in the meta-turbidites in this study (Section 4.1.1).In igneous rocks, chlorite is usually associated with hydrothermal alteration; however, it can also form during regional metamorphism (Shabani, 2009).The ISB meta-turbidites are amphibolite facies rocks and therefore are not expected to contain metamorphic chlorite that typically reflects greenschist metamorphic grade; therefore, the presence of chlorite indicates that it is a secondary feature that was most likely introduced by fluids (Chipman, 1989).The addition of hydrothermal fluid has been linked to retrogression and formation of chlorite in metamorphic complexes, so chlorite is potentially an indicator of hydrothermal alteration for the ISB (Cao et al., 2017) that it is also secondary.The appearance of sulfides filling vein structures and their recrystallized morphologies could be a sign that sulfur has been added to this sample from an outside source (although it is possible that the vein formed from remobilization of primary sulfur that was already indigenous to the rock).Lastly, the presence of chalcopyrite (CuFeS 2 ) is typically associated with hydrothermal fluids rich in copper (Haldar, 2017).Both the metaturbidites and amphibolites appear to contain molybdenum or copper sulfides suggesting that these rocks experienced hydrothermal input either during or after deposition.Shales at this locality display a correlation between Cu and total sulfur abundance (Stüeken et al., 2021), and the presence of chalcopyrite and molybdenite at the meta-turbidite locality is supported by previous mapping surveys of the outcrop (Rosing, 1999).
These observations collectively raise several potential issues with the measured sulfur isotope signatures.The Re-Os dates obtained from the molybdenite vein provide a constraint on at least some of the hydrothermal history of these rocks.We obtained two ages: 3695 ± 234 Ma for the interior of the vein and 2668 ± 295 Ma for the edge (Figure 7c).The former age overlaps with the estimated  -Salazar et al., 2021).The Neoarchean age agrees with a recently identified second metamorphic event (Eskesen et al., 2023) and likely represents partial recrystallization of the Eoarchean molybdenite.Metamorphic recrystallization is also evidenced from the textures (Figures 2 and 3).In the case of the meta-carbonates, we were not able to date sulfides, but the Rb-Sr ages of the mica (Figure 8d) agree with the Proterozoic ages of shear zones in the Mesoarchean Akia terrane of Greenland (Kirkland et al., 2023).This opens the possibility of metasomatic alteration associated with shearing at a post-Archean time.and the Ameralik Dykes in the IGC (0.02‰) (Mojzsis et al., 2003;Ueno et al., 2008), indicating a magmatic sulfur source with ∆ 33 S of ~0‰.This contrasts with the larger ∆ 33 S values seen in the metaturbidites and meta-carbonates.Collectively, our petrographic observations and isotopic results may thus indicate the presence of a primary MIF signal that was overprinted and perhaps somewhat muted by Archean and/or Proterozoic fluids that did not contain MIF.Further analyses would be needed to fully discount the possibility that some sulfur phases were introduced during Neoarchean or Palaeoproterozoic metamorphism (Eskesen et al., 2023).

| CON CLUS I ON AND IMPLI C ATI ON S
To conclude, our sulfur isotope analysis of whole-rock samples by MC-ICP-MS is both accurate and precise.This method can,  Eskesen et al., 2023).The chalcopyrite remains undated.We therefore cannot rule out the possibility that these hydrothermal fluids also introduced the MIF signal, mobilized from younger Archean sedimentary strata; however, a primary origin remains possible.
The ISB meta-carbonates exhibit a negative MIF signature of a similar magnitude to previously measured Archean barites.However, the aqua regia digestion method does not fully attack barite, and sulfides were more prevalent in the samples analyzed.The most likely host phase of sulfur in these rocks was pyrite.Its negative Δ 33 S signature is most parsimoniously explained by the reduction of marine sulfate, similar to what has been described from slightly younger Paleoarchean rocks (Roerdink et al., 2013;Ueno et al., 2008).Nearby 14724669, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gbi.12595by Test, Wiley Online Library on [10/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License More recently, Schurr et al. (2020) compared MC-ICP-MS to EA-IRMS for δ 34 S measurements of calcium-sulfate minerals and CAS, finding good agreement with two orders of magnitude less sample for MC-ICP-MS.This technique therefore holds great potential for unlocking sulfur isotope records of other Archean archives, especially for researchers that may not have access to fluorination equipment or a NanoSIMS, which are still relatively rare instruments.Here we developed the MC-ICP-MS technique with

F
I G U R E 1 (a) Location of Isua within the North Atlantic craton in SW Greenland.Modified from Kolb et al. (2015).(b) Map of the ISB and surrounding area.(c) Geological map of sample area on the western side of the ISB.Modified from δ 33 S ∕ 1000 − δ 34 S∕1000 + 1 0.515 − 1 14724669, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gbi.12595by Test, Wiley Online Library on [10/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License maps show the distribution of elements and thereby textures and mineral phases can be investigated.The element maps are presented as two-dimensional image files, however, because the fluorescence X-rays come from different depths, there are some interferences in the signal that require interpretation to deduce the mineralogy at the level of each pixel.

A
Bruker D8 Advance diffractometer was used for powder x-ray diffraction (PXRD) analyses at the University of Copenhagen.The instrument has a primary Ge111 monochromator and a LinxEye silicon-strip detector with an active surface covering 3.3°.The used radiation was from a sealed Cu tube (1.54059 Å).The measurements were made between 5° and 90° 2θ, in steps of 0.02° and a measuring time of 4 s.The reflection Bragg-Brentano technique was used with sample rotation and a fixed divergence at 0.25°.Sample preparation included crushing samples to a grain size of <45 μm and mounting the powder in a metal sample holder with a cavity of 2 mm depth.Mineral identification was done using Bruker's EVA software (DIFFRAC.EVA version 7), and the associated database with mineral diffraction data.Semi-quantitative phase analysis was achieved using curve fitting within this software.
14724669, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gbi.12595by Test, Wiley Online Library on [10/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licenseabundance.The laser fluence was set at 3.5 J cm −2 with a repetition rate of 5 Hz and a spot-size of 67 μm.Analyses included 30 s of background collection followed by 40 s of ablation.Data reduction was conducted in the LADR software(Norris & Danyushevsky, 2018) Meta-turbidite 208293 is the most highly altered of the four samples.The section overall has a cloudy and turbid texture indicative of alteration.It is predominantly composed of quartz and biotite with no replacement by chlorite as seen in other samples.Unlike the other meta-turbidite samples, 208293 contains zoisite and epidote group minerals which is typically associated with rocks that have undergone medium-grade regional metamorphism(Pichler & Schmitt- Riegraf, 1997).The sample also contains patches of likely secondary carbonate.No visible sulfides are seen under reflected light, and graphite is sparse.4.1.2| ISB meta-carbonatesMeta-carbonate samples 207850(1) and 207850(2) contain abundant carbonate, quartz, and biotite with small amounts of accessory opaques (Figure3a,b).In line withNutman et al. (2016), there is no evidence of tremolite.Sample 207850(2) contains minor amounts of pyrite, visible under reflected light and identifiable by its color and cubic crystal shape (Figure3c,d).No obvious evidence for sulfate minerals such as barite and anhydrite was seen.

TA B L E 2
Sulfur isotope data for ISB samples.14724669, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gbi.12595by Test, Wiley Online Library on [10/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licenseprecision of this method.Similarly, the Proterozoic shale samples fell within the error range for previously measured values(Nielson et al., 2024; Table

F I G U R E 7
Re-Os geochronology data for reference materials and the Isua molybdenite vein.(a, b) Illustrate analytical reproducibility for two reference materials.(c) Shows the measured ages for the interior (3695 ± 234 Ma, MSWD = 0.02) and edge of the molybdenite vein (2668 ± 295 Ma, MSWD = 0.05).(d) Shows and alternative isochron regression through all data, excluding the three younger outliers, anchored to a present-day initial 188 Os/ 187 Os ratio set at 6.74 ± 1.00, which produced a Re-Os date of 3662 ± 214 Ma. depositional age of the sediments and so may indicate the syngenetic introduction of hydrothermal fluids.This interpretation opens the exciting possibility of a hydrothermal Mo source into the Eoarchean ocean that may have been beneficial for early life (Schoepp-Cothenet et al., 2012).Alternatively, the molybdenite, including the sulfur, may have been introduced by metamorphism shortly after sediment deposition, consistent with the metamorphic history of the region (Ramírez If the hydrothermal fluid carried an MDF signature (i.e., no MIF), as one would expect for a magmatic or mantle-derived fluid, this would mix with the existing rock and act to weaken the MIF signal preserved in Archean sediments toward ~0‰ ∆ 33 S values(Papineau & Mojzsis, 2006).In contrast, a marine hydrothermal fluid from the Archean ocean would likely have the negative ∆ 33 S values that has been proposed for Archean seawater sulfate(Baumgartner et al., 2020) and may thus push sedimentary ∆ 33 S to negative values.If instead the fluid was derived from the post-Archean ocean, its ∆ 33 S composition would again have been zero and diluted any non-zero ∆ 33 S signatures.Lastly, it is possible that the hydrothermal fluid mobilized sedimentary sulfur from younger Archean rocks with a positive or negative MIF signature and deposited that sulfur in these Eoarchean strata, thereby introducing MIF into the Eoarchean strata analyzed in this study.We cannot discount any of these possibilities and highlight that future in situ analyses would help to distinguish different generations of sulfides and potentially rule out different sources of sulfur F I G U R E 8 Rb-Sr geochronology data for reference materials and meta-carbonate samples.(a) MDC phlogopite used for calibration; (b) Bund-1b quality control reference material; (c) Taratap quality control material; (d) results for Isua meta-carbonate sample.inthese rocks.Nevertheless, there are reasons to believe that the meta-sedimentary rocks display primary MIF.First, the Eoarchean Re-Os age derived from the molybdenite suggests that at least some hydrothermal overprinting occurred during or shortly after sediment deposition.The molybdenum was likely sourced from the leaching of magmatic rocks, as marine and sedimentary Mo levels were very low at that time(Johnson et al., 2021).Hence the ∆ 33 S signature of this magmatic fluid would have been close to a magmatic value of zero per mil, and, therefore, the presence of MIF in the meta-turbidite is perhaps best explained by primary MIFcarrying sulfur in the sediments.Second, the amphibolite sample 207849C displayed significant veining and secondary sulfide emplacement, and yet the ∆ 33 S value of this sample falls just outside but within error of the MDF threshold with a ∆ 33 S value of 0.16‰.Both our amphibolite ∆ 33 S values are similar to the composition of previously measured amphibolites from the ISB (−0.17‰ to 0.26‰, average of 0.02‰,Siedenberg et al., 2016) and other Archean basalt from the Dresser Formation in western Australia (−0.100‰) therefore, become an important new addition to the geochemical toolbox as it opens up analyses of multiple sulfur isotopes in rock samples with very low sulfur content where NanoSIMS or miniaturized fluorination techniques are not available.Although this method cannot distinguish between individual sulfide phases, the distinct patterns that we identified between the different localities indicate that it can nevertheless provide important new insights into the predominant sulfur sources in a given sample set.ISB meta-turbidites exhibit a positive MIF signature characteristic of Archean sulfides formed in a low oxygen atmosphere, in line with previous analyses of schists and BIFs in the ISB.Pyrite, as well as chalcopyrite and in one case molybdenite, were likely the main sulfur host mineral in these samples.The presence of chalcopyrite and molybdenite, along with textural observations, suggests the input of hydrothermal fluids, either during or after sediment deposition.The Re-Os date obtained from the molybdenite points to at least one Eoarchean fluid injection event, followed by recrystallization during the Neoarchean (cf. amphibolites, despite evidence of secondary fluid infiltration, do not show significant MIF, suggesting that at least the secondary sulfur in those samples had a magmatic source.New Rb-Sr data reveal a Proterozoic age for micas in the meta-carbonate, potentially F I G U R E 9 (a) Plot of δ 34 S vs. ∆ 33 S for this study compared with sulfate and sulfide measurements through Earth history, taken from the compilation by Claire et al. (2014).ARA, Archean Reference Array; MDF, Mass Dependent Fractionation.(b) Compilation of ∆ 33 S values for sulfates, sulfides and CAS vs. time.