Multiple Sulfur Isotope Evidence for Bacterial Sulfate Reduction and Sulfate Disproportionation Operated in Mesoarchaean Rocks of the Karelian Craton

Abstract

Sulfur isotope in sulfides from the Paleoarchean and the Neoarchean sedimentary rocks evidence microbial sulfur metabolism in Archean sulfur cycle. However, sulfur metabolism for the Mesoarchean interval is less obvious since evidence for a large range in sulfur isotope values has not yet been observed in Mesoarchean samples. We report the results of multiple sulfur isotope measurements for sulfide minerals from ~2.8 Ga sedimentary rocks in the southeastern part of the Karelian Craton. In situ isotope analysis of sulfide grains have been performed using a femtosecond laser-ablation fluorination method. Sulfide samples studied here yielded Δ³³S values between −0.3 and +2.7‰ and δ³⁴S values between −10 and +33‰. The Δ³³S dataset was interpreted to indicate the incorporation of sulfur from two coexisting sulfur pools, photolytic sulfate and photolytically derived elemental sulfur. We suggest that the relative contributions of these Δ³³S different pools to the pyritic sulfur could be controlled by the metabolic activity of coexisting sulfate-reducing and sulfur-disproportionating bacteria during pyrite formation. We therefore suggest the operation of different metabolic pathways of sulfur in Mesoarchean sedimentary environments.

1. Introduction

Bacteria are widespread prokaryotic organisms observed in different modern environments [1,2] and are thought to have already inhabited the Earth in the Paleoarchean [3]. Microbial activity in Paleoarchean (3.6–3.2 Ga) sedimentary environments has been identified on the basis of a combination of mass-dependent and mass-independent fractionation of sulfur isotopes (δ³⁴S and Δ³³S values) in sulfides from the Dresser Formation (3.49 Ga) in the Pilbara Craton, Western Australia [4,5,6,7,8,9] and from the Barberton Greenstone Belt, South Africa [5,10,11,12]. The strongly negative δ³⁴S values recorded in sulfides from ancient sediments have been considered as one of the principal arguments for a biological genesis of sulfides [13,14], whereas reactions of thermochemical sulfate reduction have been proposed as an alternative for the process of pyrite formation with negative δ³⁴S values in the Dresser Formation [15].

Studies of sulfide minerals from Neoarchean (2.8–2.5 Ga) rocks of Western Australia, South Africa and Brazil [16,17,18,19,20,21] have also shown large sulfur isotope fractionations (collectively, δ³⁴S data published in these studies range from values of approximately −30‰ to +20‰), that supports the existence of microbial life during the Neoarchean.

In contrast to the relatively wide range of variability for δ³⁴S values in Paleoarchean and Neoarchean sediments, sulfur isotope data that are currently available in the literature for Mesoarchean (3.2–2.8 Ga) sediments have shown a significantly smaller range in δ³⁴S values. For example, Mesoarchean samples from southern Africa and Western Australia yielded δ³⁴S values that predominantly grouped within a range from −3‰ to +7‰ [22,23,24]. Based on the lack of significant fractionations in ³⁴S that is expected in microbially driven processes, Ono and coauthors [24] considered microbial sulfur cycling to be a minor factor for the transfer of sulfur species from the oceanic reservoir into sulfide sulfur in sedimentary rocks of the ~2.9 Ga Mozaan Group of the Pongola Supergroup in Southern Africa. The lack of evidence for large sulfur isotope fractionations has been linked to the limited bacterial sulfur cycling in ancient pyrite-forming environments [25]. However, despite the smaller δ³⁴S variations in Mesoarchean compared to other Archaean intervals, there is still uncertainty with regard to the presence of microorganisms in Mesoarchean oceans since a narrow range in δ³⁴S values in sulfides does not yet indicate the absence of biogenic transformations of sulfur compounds, but there is no evidence of active microbial activity, i.e., observations of significant fractionation in ³⁴S for sedimentary rocks, to trace bacterial cycling of sulfur in the Mesoarchean. Thus, further research is needed to gain clarity on this issue.

There is now a unique opportunity to identify different microbially driven processes and sulfur cycling on the early Earth due to multiple sulfur isotope (³²S, ³³S, ³⁴S, and ³⁶S) measurements and the existence of mass-independently fractionated sulfur isotope signatures (non-zero Δ³³S and Δ³⁶S values) in sediments older than 2.4 Ga [26,27]. The presence of non-zero values for Δ³³S and Δ³⁶S in ancient sedimentary rocks may reflect photochemical gas-phase reactions of volcanogenic sulfur dioxide (SO₂) in an anoxic atmosphere [28,29], which were responsible for the formation of a negative Δ³³S-reservoir of sulfate and a positive Δ³³S-reservoir of elemental sulfur [30]. The observations of negative and/or positive Δ³³S values in Archean sedimentary pyrite may be helpful for identifying different sources of sulfur that would have been involved in microbially driven processes. Revealing the correlation between Δ³³S and Δ³⁶S values may be especially helpful to make a differentiation between mass-dependent and mass-independent processes [31].

In this contribution, we present a study of sulfur isotope composition (δ³⁴S, Δ³³S and Δ³⁶S) of sulfides from the Mesoarchean volcanic-hosted massive sulfide occurrences in the southeastern part of the Karelian Craton by using laser-ablation sulfur isotope analyses. This study was performed in order to elucidate the sources of sulfur and the possible participation of bacteria in the transformation of sulfur to sulfides for occurrences of Mesoarchean age. Isotope data obtained in our work and their interpretation unequivocally indicate that microbially driven processes operated in Mesoarchean sedimentary environments, including two types of microbial sulfur metabolism–sulfate reduction and sulfur disproportionation.

2. Geological Setting

The Karelian Craton (Figure 1A) is the largest block of the Archean crust of the Fennoscandian (Baltic) Shield. It is formed by typical Archean granite–greenstone association and consist of granite–gneiss, greenstone, paragneiss and granulite complexes [32,33,34,35,36].

Three subprovinces are distinguished within the Craton: Vodlozero, Central Karelia and West Karelia, each possessing unique development characteristics during the Archean [35,37,38].

The Paleo- (3.3–3.1 Ga) and Meso-Archean tonalite-trondhjemite-granodiorite (TTG) gneisses and Mesoarchean (3.05–2.8 Ga) greenstone complex formed Vodlozero subprovince [34,42]. The Sumozero-Kenozero greenstone belt is located in the eastern part of it. It is one of the largest greenstone belts of the subprovince. It extends for ~400 km and is up to 50 km in width [34,39,41]. This greenstone belt is a system of the tectonic sheets and consists of slightly metamorphosed volcanic, volcanic-sedimentary, and intrusive rocks. Paleoproterozoic basaltic overlies the Archean sequence [35,41].

The Kamennoozero belt is one of the Sumozero-Kenozero greenstone belt fragments (Figure 1B). It includes two main units (tectonic sheets) with a total thickness of ~5 km [34,39]. The lower unit consists of pillow metabasalts and komatiite showing an affinity to oceanic plateau volcanism. The ages of this lower unit are 2892 ± 130 (by Pb-Pb) and 2916 ± 117 Ma (by Sm-Nd) [39]. The upper unit consists of volcanic basalt–andesite–dacite-rhyolite (BADR)-series rocks, bearing arc signatures, and adelite-series subvolcanic rhyolites. This volcanic rock age is 2875 ± 2 Ma (by U-Pb zircon TIMS) [39]. The unique feature of the upper unit is the presence of sulfide ores [40,43,44], which served as the object of this study.

At the end of the last century, over 10 sulfide deposits were explored in the Kamennoozero belt by exploratory drilling (see Figure 1B). The volcano-sedimentary massive sulfide (VSMS) Leksa deposit studied here is located in the southeastern part of the Kamennoozero belt and occurs in strata of quartz–albite–sericite and shungite-bearing schists [40].

The carbon-bearing schists are made of thin veinlets and lenses of quartz and ferruginous chlorite in a fine-grained matrix built by a mixture of chlorite, quartz, muscovite, and shungite. Quartz–albite–sericite schists are metasomatic rocks, presumably former volcanogenic-sedimentary arc-type rocks [40]. Their mineral composition varies from carbonate–quartz–albite–sericite–chlorite to quartz–pyrite. Carbonate is found almost everywhere, which is characteristic of carbon-dioxide metasomatism.

Sulfide ores constitute several superimposed ore bodies (Figure 2) that formed two main sulfide-bearing sequences totally up to 40 m thick. Host rocks near the ore body underwent an intensive silicification. Sulfides are represented mainly as pyrite forming different morphological shapes (cubic crystals, framboids, rounded globules, and layered concretions), rarer as pyrrhotite, chalcopyrite, sphalerite, and galena. Ores are impregnated, spotted, banded, massive, sometimes with veinlets, and of mainly pyritic composition 30%–50% of which are sulfides. The concentration of polymetals in ores is low (Zn 0.03%–0.24%; Au 0.003–0.06 ppm, Ag up to 0.02–4.00 ppm) [45]. The mineralization is characterized by being stratiform and associated with island-arc sedimentary-volcanogenic felsic rocks. Mafic lavas were not found by drilling, although field observations and the presence of fuchsite in metasomatites indicate their presence somewhere nearby.

Layered or banded, disseminated-banded and disseminated-streaky ore textures are usually typical for distal (distant from the hydrothermal discharge zone) facies of volcanogenic-sedimentary massive deposits [46]. Considering the established metal zonation at many VSMS deposits [47,48], the wells at the Leksa deposit possibly revealed rather low-temperature sulfide ore facies distant from the discharge hydrothermal zone. Based on the lithological classification of the host rock [49,50], we have classified this deposit as bimodal-mafic VSMS. Volcanic rocks of a VSMS deposit of this type are dominated by basalt, andesite, felsic lavas, and pyroclastic rocks. These deposits usually indicate hydrothermal fields related to island-arc volcanoes and backarc spreading centers [51].

3. Materials and Methods

Samples were collected from three drill cores: H1, H2, and H3. The cores intersect highly carbonaceous and carbonate–quartz–sericite shales. The choice of samples for the study was guided by the purpose of this research, which is to trace potential sulfur isotope signatures of Mesoarchean life. Therefore, sulfide samples were taken mostly from metasedimentary lenses in ore bodies of the Leksa deposit.

Two textural types of pre-metamorphic pyrite have been identified in these shales. Detailed sample descriptions are in Appendix, Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5. One type is represented by aggregates of nano-micro crystals with different sizes, for which a generic term could be colloform-like pyrite. Colloform pyrite occurs as ellipsoidal aggregates composed of densely packed submicron-sized crystals and also as concentrically laminated grains. They are common in both clay-carbonaceous and carbonate shales. An example of concentrically laminated grains presented in Figure 3A: pyrite layers alternate with layers of marcasite; interstitial spaces of the marcasite crystals are filled with chlorite (chamosite), sericite (muscovite), quartz crystals and graphite. Note that since our study does not address issues related to the conditions favorable for the formation of pyrite or marcasite, in further discussion of the results we use the term pyrite to refer both to pyrite and its polymorph, marcasite.

Another type of pyrite (Figure 3B–D) studied here is represented by euhedral pyrite crystals. They commonly occur as individual microcrystals disseminated in the black carbonaceous shale (Figure 3B), or as thin layers composed of pyrite grains with a size of <100 μm. They also occur as aggregates of a few crystals with a size of ~0.5 mm that have overgrowth texture around concentrically laminated pyrite grains (Figure 3A). In some cases, pyrites were observed as ring-shaped and ball-shaped aggregates (typically less than 20 μm in diameter) of microcrystals, surrounding the inner core comprised of silicate minerals: quartz, chlorite, muscovite, or their mixture (Figure 3C,D).

From pyrite-bearing rock fragments of the drill-cores, polished chips (approximately 1–3 cm length/width and 0.5 cm deep) were prepared for both petrographic observation and isotope analyses. Samples used for isotope analysis were not subjected to additional processing, including etching in HNO₃. All analyses were performed at Far East Geological Institute FEB RAS in Russia. More details on sulfide petrography are provided in the Appendix A.

3.1. Sulfur Isotope Analysis

Femtosecond laser-ablation fluorination method described in detail in [52,53] was applied for in situ measurements of δ³³S and δ³⁴S values in sulfide minerals. Briefly, an ultraviolet femtosecond laser ablation system (NWR Femtosecond UC with laser Pharos 2mJ-200-PP and harmonics module HE-4Hi-A, supplied by Electro Scientific Industries New Wave Research Division, Portland, OR, USA) was used to produce a laser crater of approximately 80 μm diameter and 40 μm depth in pyrite. Laser-generated aerosol sulfide particles were converted to SF₆ (~12–13 nmol) via reaction with BrF₅ at 350 °C. The SF₆ was purified by means of a cryogenic trap system [53] and introduced into the ion source of a Thermo Scientific MAT 253 isotope ratio mass spectrometer using a gas injection interface developed and described in [52]. Sulfur isotope ratios were measured by monitoring the SF₅+ ion currents at mass to charge ratios of m/z 127 (³²SF₅⁺), 128 (³³SF₅⁺), 129 (³⁴SF₅⁺), and 131(³⁶SF₅⁺).

It should be noted that it is not a problem if the size of the ablation pit is larger than the size of the analyzed sulfide grain because the presence of the matrix rock did not affect the accuracy and precision of analyses [53].

All sulfur isotope data are presented as conventional notation:

$$\mathsf{\delta }=\frac{{\mathrm{R}}_{\mathrm{sample}}}{{\mathrm{R}}_{\mathrm{standard}}}-1,$$

where R_sample and R_standard are the ³⁴S/³²S or ³³S/³²S ratios for the sample and standard, respectively, and reported relative to Vienna Cañon Diablo Troilite (V-CDT). For describing sulfur isotope anomalies (mass-independent sulfur isotope fractionations), capital delta notation was used which is defined as follows [54]:

$${\Delta }^{33}\mathrm{S}={\mathsf{\delta }}^{33}\mathrm{S}-\left[{\left({\mathsf{\delta }}^{34}+1\right)}^{0.515}-1\right]$$

$${\Delta }^{36}\mathrm{S}={\mathsf{\delta }}^{36}\mathrm{S}-\left[{\left({\mathsf{\delta }}^{34}+1\right)}^{1.90}-1\right]$$

Three International Atomic Energy Agency standards (IAEA S1, S2, and S3) were used to calibrate the SF₆ reference gas and in-house standard. Analytical precision of ±0.2‰ (1σ) for δ³⁴S values, ±0.03‰ (1σ) for Δ³³S values, and ±0.27‰ (1σ) for Δ³⁶S values were estimated from the long-term (approximately 1-year-long period) reproducibility of in-house standard [53].

3.2. Scanning Electron Microscopy and Energy Dispersion X-ray Spectroscopy Analysis

• Scanning electron microscopy (SEM) was used to examine the detailed mineralogical and textural characteristics of sulfides on polished and some unpolished surface areas. Backscattered electron (BSE) and secondary electron (SE) imagery, as well as qualitative energy dispersive X-ray spectroscopy (EDS) analysis, was carried out on gold or carbon coated samples using a dual-beam TESCAN LYRA 3 XMH (Schottky cathode) Oxford AZtec Energy EDS system.
• Electron imagery was performed at variable acceleration voltage (20–30 kV) and beam current (9–14 nA). Up to 30 kV and 14 nA was used for EDS analysis to ensure sufficiently high peak count rates for accurate determination of characteristic element-specific X-ray emission lines.
• Quantitative X-ray spectroscopy spot analysis was performed using a JEOL JXA 8100 electron probe micro-analyser with three wave spectrometers and one energy dispersive spectrometers (Oxford Instruments Inca, Abingdon, UK), under a resolution of 137 eV MnKα, an accelerating voltage of 20 kV and a measure current of 1 × 10⁻⁸ A. Prior to analysis the samples were coated with a 20 nm carbon film. The beam was fully focused to give a spot size of about 1 μm with a measure current 1 × 10⁻⁸ A. Pure metals, glasses, and minerals analyzed through other methods were used as standards along with Oxford Instruments standards. Total Fe is equivalent to Fe⁺² in calculations.
• The qualitative phase analysis of sulfides was determined by X-ray diffraction in the powder by Rigaku MiniFlex II (Rigaku, Japan) (XRD) at the Laboratory of X-Ray Methods of the Analytical Center of the Far East Geological Institute, Far Eastern Branch of the Russian Academy of Sciences (FEGI FEB RAS) in Vladivostok.

4. Results

Sulfur isotope data are presented in Appendix A,

Table A1. Isotopic data for pyrite from sedimentary rocks of the Leksa deposit.

Sample (Drill Core/Depth)	δ34S (‰)	Δ33S (‰)	Δ36S (‰)	Pyrite Types
H1/50.7	−1.3	0.03		Euhedral crystal disseminated in rocks
H1/50.7	−0.1	−0.04		Euhedral crystal disseminated in rocks
H1/50.7	−0.9	−0.02		Euhedral crystal disseminated in rocks
H1/50.7	−1.2	0.03		Euhedral crystal disseminated in rocks
H1/50.7	5.4	0.04		Euhedral crystal disseminated in rocks
H1/54.6	13.4	0.02		Euhedral crystal disseminated in rocks
H1/54.6	13.9	0.34		Euhedral crystal disseminated in rocks
H1/54.6	13.5	−0.02		Euhedral crystal disseminated in rocks
H1/54.6	−3.2	0.00		Euhedral crystal disseminated in rocks
H1/54.6	−2.9	−0.05		Euhedral crystal disseminated in rocks
H1/63	−1.0	0.00		Euhedral crystal disseminated in rocks
H1/75	−4.8	0.84		Euhedral crystal disseminated in rocks
H1/75	−4.2	−0.38		Euhedral crystal disseminated in rocks
H1/75	−7.0	0.23		Euhedral crystal disseminated in rocks
H1/75	−1.0	−0.51		Euhedral crystal disseminated in rocks
H2/24.0	31.9	−0.18		Euhedral crystal disseminated in rocks
H2/24.0	32.7	−0.16		Euhedral crystal disseminated in rocks
H2/61.5	2.6	−0.01		Euhedral crystal disseminated in rocks
H2/61.5	2.3	0.04		Euhedral crystal disseminated in rocks
H2/61.5	−5.1	0.05		Euhedral crystal disseminated in rocks
H2/61.5	2.4	−0.02		Euhedral crystal disseminated in rocks
H2/61.5	2.2	0.02		Euhedral crystal disseminated in rocks
H3/56.0	13.8	1.48		Euhedral crystal disseminated in rocks
H3/56.0	12.6	1.28		Euhedral crystal disseminated in rocks
H3/56.0	7.6	0.25		Euhedral crystal disseminated in rocks
H3/56.0	11.5	0.40		Euhedral crystal disseminated in rocks
H3/56.0	13.8	0.12		Euhedral crystal disseminated in rocks
H3/56.0	11.6	0.28		Euhedral crystal disseminated in rocks
H3/56.0	12.1	0.32		Euhedral crystal disseminated in rocks
H3/56.0	10.2	0.28		Euhedral crystal disseminated in rocks
H3/56.0	12.0	0.35		Euhedral crystal disseminated in rocks
H3/56.0	11.9	0.28		Euhedral crystal disseminated in rocks
H3/56.0	10.3	0.41		Euhedral crystal disseminated in rocks
H3/56.0	5.4	0.10		Euhedral crystal disseminated in rocks
H3/56.0	16.9	0.67		Euhedral crystal disseminated in rocks
H3/56.0	12.2	0.42		Euhedral crystal disseminated in rocks
H1/75	9.9	1.65		Euhedral, observed as ring-shaped aggregates
H1/75	11.0	2.65		Euhedral, observed as ring-shaped aggregates
H1/75	3.7	1.76	−1.3	Euhedral, observed as ring-shaped aggregates
H1/75	2.8	1.57	−2.0	Euhedral, observed as ring-shaped aggregates
H1/75	3.8	1.93	−2.4	Euhedral, observed as ring-shaped aggregates
H1/75	3.8	1.85	−2.5	Euhedral, observed as ring-shaped aggregates
H1/75	6.1	1.95		Euhedral crystal, overgrowth around colloform grains
H1/75	6.3	2.20		Euhedral crystal, overgrowth around colloform grains
H1/75	5.2	1.55		Euhedral crystal, overgrowth around colloform grains
H1/75	4.6	1.55		Euhedral crystal, overgrowth around colloform grains
H1/75	5.3	1.93		Euhedral crystal, overgrowth around colloform grains
H1/75	5.2	1.96		Euhedral crystal, overgrowth around colloform grains
H1/75	5.1	1.85		Euhedral crystal, overgrowth around colloform grains
H1/75	7.0	2.64		Euhedral crystal, overgrowth around colloform grains
H1/75	5.7	2.06		Euhedral crystal, overgrowth around colloform grains
H1/75	5.9	2.02		Euhedral crystal, overgrowth around colloform grains
H1/75	7.8	1.40		Colloform grains_1, core
H1/75	3.7	0.31		Colloform grains_1, concentric layer
H1/75	0.5	0.00		Colloform grains_1, concentric layer
H1/75	−1.8	0.00		Colloform grains_1, concentric layer
H1/75	3.0	−0.01		Colloform grains_1, concentric layer
H1/75	16.2	−0.01		Colloform grains_1, concentric layer
H1/75	27.5	0.44		Colloform grains_1, concentric layer
H1/75	24.5	0.33		Colloform grains_1, concentric layer
H1/75	15.4	−0.01		Colloform grains_1, concentric layer
H1/75	−6.7	0.05		Colloform grains_1, concentric layer
H1/75	−9.5	0.13		Colloform grains_2, non-laminated
H1/75	−9.8	0.22		Colloform grains_2, non-laminated
H1/75	−10.2	0.11		Colloform grains_2, non-laminated
H1/75	7.0	0.08		Colloform grains_3, non-laminated
H1/75	1.6	−0.30		Colloform grains_3, non-laminated
H1/75	−6.6	−0.27		Colloform grains_3, non-laminated
H3/56.0	5.7	0.17		Colloform grains_4, core
H3/56.0	8.5	0.23		Colloform grains_4, concentric layer
H3/56.0	10.6	0.48		Colloform grains_4, concentric layer
H3/56.0	10.1	0.49		Colloform grains_4, concentric layer
H3/56.0	11.9	0.45		Colloform grains_4, concentric layer

and displayed in Figure 4 as plots of δ³⁴S vs. Δ³³S values.

Colloform pyrite shows both negative and positive δ³⁴S values; a total δ³⁴S range is from −10.2‰ to +27.5‰ (Figure 4). In situ analyses revealed strongly heterogenous δ³⁴S values, ranging from −6.7‰ to +27.5‰ in concentrically laminated pyrite grains, and relatively homogenous δ³⁴S values that range from −10.2‰ to −9.5‰ and from −6.6‰ to +7.0‰ in non-laminated pyrite grains (Appendix A, Table A1). It is also evident that colloform pyrite studied here is mass-independently fractionated, with Δ³³S values that mainly vary between −0.3‰ and +0.5‰. The exception is relatively large positive Δ³³S = +1.4‰ measured in the core of the concentrically laminated pyrite (see Appendix A, Figure A4 and Figure A5).

The euhedral pyrite crystals have showed a wide spread of δ³⁴S values and also revealed the presence of mass-independently fractionated sulfur. More specifically, disseminated pyrite exhibited intergranular heterogeneity for δ³⁴S and Δ³³S values that vary between −7.0‰ and +32.7‰ and between −0.5‰ and +1.5‰, respectively (Figure 4). These data are similar to those obtained for concentrically laminated pyrite grains. In contrast, euhedral pyrite overgrowths observed around colloform pyrite are relatively homogeneous in δ³⁴S and Δ³³S values that vary between +4.6‰ and +7.0‰ and between +1.5‰ and +2.7‰, respectively (Figure 4). Two analysis of euhedral pyrite observed as ring-shaped and ball-shaped aggregates of microcrystals have showed δ³⁴S = +11.0‰ and +9.9‰, Δ³³S = +2.7‰ and +1.7‰ (Table A1, Figure 4).

A subset of the euhedral pyrite samples was measured for ³⁶S/³²S isotope ratios to understand the relationship between Δ³³S and Δ³⁶S values. These pyrite grains have negative Δ³⁶S values which range from −1.3 to −2.5‰ (Table A1), and also show a negative correlation between Δ³³S and Δ³⁶S with a slope of ≈−1.15 (Figure 5).

5. Discussion

As pointed out earlier in the introduction, the range of variability for δ³⁴S values of sulfides from the Mesoarchean rocks represented by sedimentary units in southern Africa [22,23,24] and Western Australia [22] is significantly smaller compared to other Archaean intervals. Sulfur isotope data reported in this study, however, reveal a much wider range of δ³⁴S values (−10.2‰ to +32.7‰) in the 2.8 Ga old rocks of the Karelian Craton (Figure 6). Figure 4 also illustrates one feature of the sulfur isotope record identified in our study, which is the presence of relatively high ∆³³S values between +1.5‰ and +2.7‰ in the euhedral pyrite grains and the relatively low Δ³³S values mostly between −0.30‰ and +0.44‰ for the colloform pyrite.

These observations aroused an obvious research interest regarding the possible pathways to differentiate the sulfur isotope ratios between these texturally distinct types of pyrite. Moreover, can the sulfur isotope data be explained in terms of known microbial processes and what these data tell us about the Mesoarchean microbial life? We suggest that considering what we know today about pyrite formation in modern sedimentary environments and sulfur isotope fractionation associated with microbial processes may help in answering these questions.

A possible mechanism of pyrite formation has been studied in several experimental works [55,56,57]. The formation of low-temperature sedimentary pyrite can occur in an anoxic environment and is mainly initiated by the reaction of dissolved H₂S with reactive iron to yield non-crystalline monosulfide:

$${\mathrm{Fe}}^{2+}+{\mathrm{H}}_2\mathrm{S}\to \mathrm{FeS}+2{\mathrm{H}}^{+}$$

(1)

The produced FeS is a transit product in the process of pyrite formation. There are two ways to transform FeS to pyrite: by the reaction of FeS with elemental sulfur

$$\mathrm{FeS}+{\mathrm{S}}^0\to {\mathrm{FeS}}_2,$$

(2)

or by the reaction of FeS with H₂S

$$\mathrm{FeS}+{\mathrm{H}}_2\mathrm{S}\to {\mathrm{FeS}}_2+{\mathrm{H}}_2.$$

(3)

5.1. Sulfur Pathways Recorded by the Studied Colloform Pyrite

In our work it can be expected, as a first approximation, that dissolved H₂S which is needed for pyrite formation in marine sediments could be produced primarily from sulfate dissolved in sea water by microbial metabolism that operated at the time of pyrite deposition. This is because the wide range from −6.7‰ to +27.5‰ and strong isotopic heterogeneity in δ³⁴S values observed at micrometer scale in our sample of concentrically laminated pyrite grains (Figure A4, Appendix A) could be attributed to biological sulfate reduction, as the most likely mechanism to produce very large fractionation of sulfur isotopes [13].

We note that colloform pyrite studied here by in situ isotope analysis displays both positive (up ~1.4‰) and negative (−0.30‰) ∆³³S values (see Figure 4). Distinct mass-independent sulfur isotope signals may indicate the transformations of sulfur species in the oxygen-poor atmosphere via photochemical reactions of volcanogenic SO₂ [26]. These ultraviolet-driven reactions should, according to photochemical experiments [58,59,60,61,62,63,64,65], produce sulfate aerosols with a negative ∆³³S signature and elemental sulfur aerosols with a positive ∆³³S signature. As a result of SO₂ atmospheric photochemistry, two compositionally and isotopically distinguished photolytic sulfur pools can be generated and maintained in an anoxic atmosphere. The absence of free molecular oxygen in the atmosphere prevents the oxidation of elemental sulfur to sulfate, and hence the homogenization of the mixing of different pools. An anoxic atmosphere favors the coexistence of various photolytic sulfur reservoirs and transfers the photolytic components out of the atmosphere to Earth’s surface. Once atmospherically derived elemental sulfur and sulfate rained out into aquatic environments, they can then be transformed into pyrite which inherits the isotope anomaly. Since the samples of colloform pyrite studied here archived opposite signs of the sulfur isotope anomaly, we can suggest photolytic seawater sulfate and photolytic elemental sulfur to be the principal sulfur source during precipitation of the pyrite grains with colloform texture.

According to the above consideration, isotope characteristics of colloform pyrite samples (i.e., non-zero ∆³³S values, the broad range and also strong isotopic heterogeneity in δ³⁴S values) may be associated with photolytic origin of its sulfur and would be consistent with forming pyrite via microbially mediated pathways, namely photolytic sulfate converted to H₂S through microbial sulfate reduction. However, it needs to be ascertained if unusually high δ³⁴S values of the studied pyrite can be associated solely with microbial sulfate reduction.

5.2. Colloform Pyrite—A Result of Microbial Sulfate Reduction Alone?

Based on current knowledge of sulfur isotope fractionations between sulfate and sulfide by bacterial sulfate reduction, the produced H₂S is ³²S-enriched vs. parent sulfate (i.e., δ³⁴S_sulfide < δ³⁴S_sulfate) [13,66,67,68,69,70]. If continued sulfate reduction takes place within a restricted system (in sediment pore waters), the concentration of sulfate in the system falls and Rayleigh fractionation occurs. As a result, the common sulfur isotopic trend observed for both remaining sulfate and pyrite formed is that δ³⁴S values progressively increase as sulfate is consumed by sulfate-reducing microorganisms [71,72]. Therefore, the simplest explanation for the high-δ³⁴S values (up to +27.5‰) observed in the concentrically laminated pyrite grains would be restricted-reservoir effects. In this situation, formation of these grains may take place along or below the sediment-water interface, not within the marine water column with a relatively large pool of available sulfate in comparison to sediment pore water, and thus Rayleigh fractionation is unlikely to occur. A sedimentary origin for the pyrite grains is also supported by preservation of their textural characteristics. Pyrite grains with concentric texture have been reported for Precambrian sedimentary successions and extensively reviewed [73]. This type of pyrite is interpreted to have formed in shallow-marine environments through accretion of pyrite mud [74,75], so the growing and aggregation of microcrystals from the inner core to the outer concentric layers can be explained by geometrical selection [76].

Concentrically laminated pyrite grains from sedimentary rocks of the Witwatersrand Basin in the central Kaapvaal Craton (South Africa) have been interpreted by Agangi and coauthors [77] to have formed at the water–sediment interface and their δ³⁴S values that range from −12.2‰ to +12.6‰ have been considered as a result of sulfur isotope fractionation during sulfate reduction accompanied by Rayleigh fractionation in a closed system (at constant sulfate−sulfide fractionation). According to these authors, more than 80% of available sulfate must be consumed by sulfate reducers, which may lead to formation of pyrite with the highest (+12.6‰) δ³⁴S values (under the assumption that an initial δ³⁴S of the fluid ~−1‰). In our work, concentrically laminated pyrite grains from the Leksa deposit in the eastern part of Karelian Craton have been observed to have the δ³⁴S values as high as +27.5‰, that is ~15‰ higher in comparison to Witwatersrand concentric pyrite [77]. Although one feasible explanation we propose for the strongly ³⁴S-enriched sulfides is microbial processes, it remains unclear whether such isotopic enrichments can be achieved only through microbial sulfate reduction, even in restricted pore waters.

To calculate the magnitude of the ³⁴S isotopic enrichment in sulfide during bacterial sulfate reduction undergoing Rayleigh-type distillation in a closed system, the equation from [78] was used. We made some assumptions regarding the initial conditions in our calculations. They are: (1) starting concentration of marine sulfate was taken as 200 μM, following the conclusions of [79] that “oceanic Archean sulfate concentrations were <200 μM”, although much lower sulfate levels ≈80 μM [80] and even <10 μM are also possible [21,81]; (2) the initial marine water sulfate had a δ³⁴S value of ~0‰ [82]; (3) the fractionation associated with sulfate reduction was taken as 10‰ (i.e., δ³⁴S_sulfate − δ³⁴S_sulfide ≈10‰), since the maximum depletion of ³⁴S into pyrite observed in our study is ~−10‰. If other mechanisms did not operate, the calculations show that more than 98% of the total sulfate in the system should be consumed by bacteria to explain pyrite highly enriched in ³⁴S up to ~ +27‰. In this case the residual sulfate concentration may expect to be <4 μM, which is too low to enable bacterial sulfate reduction because threshold concentrations for sulfate uptake have been detected at ~5 to 20 μM sulfate [83]. Thus, the current model does not provide an explanation for the high δ³⁴S values measured in concentrically laminated pyrite grains.

To elucidate the possible reasons for the formation of highly ³⁴S-enriched pyrite, other cases can be considered. It is possible that the sulfur isotopic composition of Archean seawater sulfate could be higher than 0‰. Then a Rayleigh distillation model will require the δ³⁴S value of seawater sulfate to be not less than ~+10‰ to generate pyrite with δ³⁴S ≈ +27‰ upon removal of 85% of sulfate by sulfate-reducing microorganisms in the hypothetically closed system. These considerations are in line with [84] that the δ³⁴S of Archean seawater sulfate may be in the range of +6‰ to +16‰. However, this is not the case for our samples since our model calculations with large sulfur isotope fractionations between hydrogen sulfide and sulfate of more than 25% should imply high sulfate concentrations of the Mesorchean ocean, but there exists no evidence to support this resulting deduction.

5.3. Colloform Pyrite: Evidence of Sulfur Disproportionation Accompanying Sulfate Reduction

Here we demonstrate that observed high δ³⁴S values in colloform pyrite can be explained by mixing two different pools of sulfate available simultaneously for sulfate-reducing bacteria at the site of pyrite formation. One of these pools was dissolved atmospherically derived photolytic sulfate, and the other pool was biologically produced sulfate, namely as a result of microbial disproportionation of atmospherically derived insoluble elemental sulfur. Microbial elemental sulfur disproportionation in marine sediments served as the source of sulfate and hydrogen sulfide in the pore water [85,86]:

$$4{\mathrm{S}}^0+4{\mathrm{H}}_2\mathrm{O}\to 3{\mathrm{H}}_2\mathrm{S}+{\mathrm{SO}}_4^{2-}+2{\mathrm{H}}^{+}$$

(4)

It was shown that bacterial sulfur disproportionation results in generating sulfide that is depleted in ³⁴S only up to 7‰ while sulfate is enriched in ³⁴S up to 30‰ relative to starting sulfur [86,87,88,89,90,91,92]. Then, the combination of microbial sulfur disproportionation and microbial sulfate reduction can be proposed to explain highly positive δ³⁴S values of pyrite studied here. If sulfate reduction takes place and sulfate derived from elemental sulfur is continuously added to the reservoir, pyrite produced from the sulfate mixture can reach a high positive δ³⁴S value even before the concentration of available sulfate in the pore water drops to a threshold of ~5 to 20 μM [83] and pyrite formation ceases. This interpretation suggests that pyrite was formed in sediments high in available elemental sulfur for microbial processes where elemental sulfur delivered from the atmosphere consistently fell to the bottom and accumulated in sediments, forming the elemental sulfur pool. Metabolism of elemental sulfur by bacteria resulted in the formation and maintenance of a biologically derived sulfate pool, which mixed with the atmospherically derived sulfate pool that contained within pore waters of the sediment. This process was accompanied by the metabolism of sulfate sulfur through sulfate-reducing bacteria to form pyrite. If so, disproportionation of atmospherically derived elemental sulfur in the sediment would act as a source of additional sulfate in the processes of continuous loss of atmospherically derived sulfate by bacteria out of the pore waters in the closed system. This would be running into the changes in δ³⁴S values that can be observed in different pyrite grains because their δ³⁴S values should be related directly to the concentration of available biologically and atmospherically derived sulfate in pore water. Pyrite exhibiting negative δ³⁴S values would be expected if a greater proportion of total sulfate reservoir would be atmospherically derived sulfate with δ³⁴S values of ~0‰. Pyrite exhibiting relatively high δ³⁴S values would be apparent if the δ³⁴S values of sulfate reservoir would have been locally increased due to metabolism by sulfate-reducing bacteria, where the loss of the sulfate in a closed system would be replenished from a pool of atmospherically derived elemental sulfur through bacterial metabolism of elemental sulfur.

It should be noted that a combination of sulfate reduction and sulfur disproportionation has been proposed to explain strong ³⁴S depletions in sedimentary sulfides [86,93]. Indeed, these processes may be responsible for formation of depleted rather than enriched ³⁴S sulfides, but only if the mechanism of sulfide oxidation into elemental sulfur is involved [86,93], implying the presence of an oxidizing environment. We consider such an environment unlikely to exist at the time of deposition of pyrite studied here, because it is inconsistent with the observed significant mass-independent isotopic signals in pyrite, generation and preservation of which requires an anoxic depositional environment [84].

If our assumption linking formation of colloform pyrite and two different metabolic pathways of sulfur is true, colloform pyrite should also be highly inhomogeneous at micrometer scale in the δ³⁴S values. This is because the magnitude sulfur isotope fractionation induced by bacteria, and hence pyrite δ³⁴S value, depends on microbial sulfate reduction rates, which in turn can be related directly to the concentration of available dissolved sulfate as well as organic matter. Therefore, in microenvironments where several biologically mediated mechanisms operate simultaneously, produced pyrite can be remarkably heterogeneous in δ³⁴S, which was observed for concentrically laminated pyrite grains studied here.

Proceeding from our assumption of a biogenic origin for pyrite genesis, the presence of two different metabolic pathways should be responsible not only for the local variations in δ³⁴S values, but also for Δ³³S values, which is the case for the samples of colloform pyrite. Concentrically laminated pyrite grains are characterized by a heterogeneous distribution of nonzero Δ³³S values as well as the presence of zero Δ³³S values.

The observed at micrometer scale changes in Δ³³S values of pyrite grains cannot be explained by microbial activity and associated Rayleigh fractionation of sulfur isotopes in restricted pore waters. It is because bacterial sulfate reduction and sulfur disproportionation cannot induce sulfur isotope mass-independent fractionation; the transformation of sulfur by these processes cannot significantly change Δ³³S of sulfur reservoirs. In other words, as bacterial sulfate reduction with Rayleigh distillation progressively occurs within a Δ³³S-bearing reservoir, the Δ³³S magnitude of the remaining sulfate reservoir remains unchanged and, thus, pyrite formed should be homogeneous in Δ³³S values and identical to the Δ³³S reservoir.

Strong heterogeneity in Δ³³S at the micrometer scale that we observe for colloform pyrite can be explained by relatively rapid alteration of Δ³³S for sulfur reservoir due to mixing in pore waters at least two sulfate reservoirs that should carry isotopic distinguishing Δ³³S signatures. In the present study it is suggested that the negative Δ³³S reservoir of sulfate available for use by sulfate-reducing microorganisms in pore water could have been diluted by positive Δ³³S sulfate, which was produced through bacterial sulfur disproportionation of atmospherically derived elemental sulfur accumulated in sediments. In this case, the relative bacterial sulfur disproportionation rates would locally influence and alter the integrated Δ³³S signature of dissolved sulfate in pore waters. Thereby a change in the Δ³³S value of local sulfate reservoir gives rise to a change in the pyrite from negative to positive Δ³³S values, including zero Δ³³S values. Microbial activity can be seen as a mixer if the amount of elemental sulfur was not a limiting factor for pyrite formation. This means that the presence of Δ³³S values close to zero within concentrically laminated pyrite grains studied here does not necessitate an origin of sulfur from magmatic inputs but may result from mixing of two Δ³³S-distinct sulfate reservoirs. It is conceivable that the simultaneous operation of two different metabolic pathways led to the erasing of the atmospheric signature in the pyrite studied here.

An additional issue with the existence of mass-independent isotopic signals in our samples arises: whether the observed Δ³³S values in the sample of colloform pyrite are attributable to the photolytic origin or it may be due to kinetic sulfur isotope fractionation associated with metabolic pathways.

According to experimental and modeling studies, the positive difference between Δ³³S values of sulfide and sulfate up to 0.2‰ are produced by microbial sulfate reduction (if only large isotope fractionations in δ³⁴S of >30‰) and slightly negative or positive shifts in Δ³³S values are associated with microbial sulfur disproportionation [31,92,94,95,96]. In our work isotopic data obtained for the grains with concentric texture are very difficult to associate with a purely kinetic sulfur isotope fractionation, since the magnitude detected for Δ³³S values is too large (>0.3‰) to account for the shift in Δ³³S due to microbial sulfate reduction. According to the above discussion, microbial sulfate reduction in combination with microbial sulfur disproportionation can explain both the highest δ³⁴S values (+27‰) and strong variability of the δ³⁴S and Δ³³S values measured in concentrically laminated pyrite grains at the micrometer scale.

5.4. Euhedral Pyrite: Evidence of Sulfur Disproportionation Accompanying Sulfate Reduction

The hypotheses presented here make it possible to understand what the reason is for a drastic increase in the Δ³³S values of up to +2.7‰ in euhedral pyrite grains developed around the colloform pyrite (see Figure 4) and why these euhedral grains are so homogeneous in both δ³⁴S and Δ³³S values. These questions deserve special consideration because euhedral pyrites were formed later than the colloform pyrite but yield higher Δ³³S values compared to colloform pyrite formed at an early stage of diagenesis. It would seem that an additional source of sulfur carrying a positive signal appeared in the system, for example, due to the opening of the system during late diagenesis. This scenario assumes infiltration of atmospherically derived sulfur carrying Δ³³S > 0‰ towards the site of pyrite formation, which is unlikely because elemental sulfur is an insoluble component of sediments. Instead of this we consider a closed system where the amount of atmospherically derived elemental sulfur being accumulated in the sediment was sufficient to form pyrite. Very distinct multiple sulfur isotopic signals of euhedral and colloform pyrite would simply be the result of the exhaustion of atmospherically derived sulfate within pore waters at the time when formation of euhedral pyrite occurred within sediments.

Based on the formation mechanism of early and late diagenetic pyrites, the sequence of processes that may be responsible for the isotopic features observed in different pyrite generations is proposed as follows:

(1) During early diagenesis, the main mechanism of pyrite formation with concentric texture could, as considered above, be closely related to the transformation of photolytic sulfates and elemental sulfur to pyritic sulfur by microbial processes, and therefore these pyrite grains showed the large range of variability for both δ³⁴S and Δ³³S values;

(2) As diagenesis continues during burial, diffusion of dissolved sulfate into marine sediments from overlying water decreases thereby limiting sulfate reducers activity. In this situation, the residual photolytic elemental sulfur could be a dominant source for the formation of pyrite via reactions (R1−R3). Pyrite crystallization by this mechanism occurs only as aggregates of microcrystals ~1 μm in diameter [55];

(3) With the burial of sediment, the processes of recrystallization and overgrowth of the microcrystals can result in the formation of large euhedral pyrites [97,98,99]. Accordingly, the euhedral pyrites issued by these processes were observed around the concentrically laminated pyrite grains. The remarkable intergranular and intragranular homogeneity in both δ³⁴S and Δ³³S values as well as large positive magnitude of Δ³³S values inherited from photolytic elemental sulfur are reasonably expected in these pyrites.

It is worthwhile saying that the strongly positive Δ³³S observed here for euhedral pyrite grains undoubtedly reflects the presence of mass-independently fractionated sulfur. Atmospheric origin of this sulfur was proven by including Δ³⁶S in the analysis. It was revealed that the values of Δ³³S linearly correlate with Δ³⁶S with a slope of ca. −1.15 (Figure 5), indicating an atmospheric source because the obtained slope differs from a slope of −6.85 (Figure A6), produced by mass-dependent microbially mediated processes [31], but is in good agreement with the Archean trend with a slope of −0.9 produced by mass-independently fractionated sulfur isotopes in Archean rocks.

Besides the concentrically laminated pyrite studied here, the single euhedral crystals or aggregates of a few crystals developed in fine-grained sedimentary rocks can also clearly indicate that sulfate-reducing and sulfur-disproportionating bacteria could have played an essential role at the time of deposition of these grains. This is because the large range of δ³⁴S values (−7.0‰ to +32.7‰) observed for the disseminated pyrite grains (see Figure 4) is consistent with the range in δ³⁴S variability from concentrically laminated pyrite (−6.7‰ to +24.5‰), and therefore could be explained by similar arguments that the formation of pyrite sulfur was closely related to the microbial processes of sulfate reduction, sulfur disproportionation, or a combination of both. The preservation of both positive and negative Δ³³S values in the euhedral pyrite crystals also implies the presence of at least two sources of sulfur for the formation of pyrite, that is, atmospherically derived sulfate carrying Δ³³S < 0‰ and elemental sulfur carrying Δ³³S > 0‰. An additional point to emphasize is that most of the disseminated pyrite grains analyzed in this work have moderate Δ³³S values ranging roughly between −0.3‰ and 0.5‰ (Figure 4). The exception is the cluster of crystals that form the ring-shaped or tubular structures (see Figure 3); they showed distinctively higher Δ³³S values of 1.65‰ and 2.65‰ (Figure 4). We associate these structures with the fossilization of microorganisms. Similar structures considered as evidence of microfossils have been described in Paleorchaean rocks of Western Australia [100], in Mesorchaean VSMS deposit in the Pilbara of Western Australia [101], in the Witwatersrand System of South Africa [102] and in Phanerozoic sediments [103]. An exhaustive review of microstructures that have been considered as Archean microfossilsis has recently been published by [104]. Experimental studies have shown that macroscopic structures of bacteria may be preserved by pyritization. This is shown by the presence of amorphous FeS on both the inner and outer surfaces of sulfate-reducing bacteria and by the immobilization of FeS around the bacterial microcolony [105]. Once FeS is formed it can transform into pyrite by the reaction of FeS with H₂S or S⁰ (reactions R2-R3).

Based on the fact that Δ³³S values measured for pyritized fossils in this study were as high as +2.65‰, we are certain that elemental sulfur was the dominant source for the processes that are responsible for the pyritization of microorganisms. This, of course, does not mean that pyritized fossils were necessarily disproportionating microorganisms, but may be considered in favor of the latter.

5.5. General Insights

In summary, the mutually complementary facts considered in this study as evidence of the integrated activity of sulfate-reducing and sulfur-disproportionating bacteria rather than sulfate reducers alone at the time of pyrite formation are the following:

(1) The similarity of the isotopic composition of microfossils (average δ³⁴S ≈ +10‰ and Δ³³S ≈ +2‰) to that measured both in the core of concentrically laminated pyrite (δ³⁴S ≈ +7.8‰ and Δ³³S ≈ +1.4‰) and in the euhedral pyrite surrounding it (average δ³⁴S ≈ +5‰ and Δ³³S ≈ +1.8‰);

(2) The similar range of δ³⁴S variations, spanning between −7.0‰ and +32.7‰, for both dispersed and concentrically laminated pyrite that is too large to be interpreted as microbial sulfate reduction.

The importance of microbial sulfur disproportionation for early Archean pyrite formation is also evidenced by the multiple sulfur isotope data of Philippot and coauthors [4]. They suggested that the microscopic sulfides with negative δ³⁴S and positive Δ³³S values in barite deposit (Dresser Formation, Western Australia) are likely due to the activity at this time of the coexisting sulfur-disproportionating, sulfur-reducing and sulfate-reducing microorganisms. However, an alternative origin for this negative relationship between Δ³³S and δ³⁴S in microscopic pyrite has been proposed by Bao and his team [106]; that is, the mixing scenario where sulfides derived from S⁰ (produced by the 193 nm SO₂ photolysis) and sulfides derived from seawater sulfate could be mixed in different proportions. Moreover, it is still possible that that the puzzle of isotopic effects will fit if sulfur-disproportionating bacteria that can also reduce sulfate were involved in the generation of the micropyrites [107].

Although the fact of the existence of microorganisms in the Archaean is well established today, less is known about the communities of bacteria which can utilize sulfur species for their metabolism in an Archaean environment. If the isotope data for sedimentary pyrite studied here indicate the coexistence of sulfur-disproportionating and sulfate-reducing microorganisms in the Mesoarchean, the sedimentary circumstances should be expected to favor the growth of both groups of these organisms. It is important that bacterial sulfur disproportionation is more sensitive to the hydrogen sulfide concentrations than bacterial sulfate reduction. In experimental studies [87,108] it was evidenced that utilization of elemental sulfur by disproportionating bacteria occurs only in the presence of a hydrogen sulfide scavenger because disproportionation of elemental sulfur is inhibited as the hydrogen sulfide concentrations raised >1 mM. In early anoxic oceanic environments with abundant Fe²⁺ in the Archean oceans [109], the presence of Fe²⁺ in pore water could act as a scavenger, buffering the hydrogen sulfide concentrations in pore water to low levels, even with rapid rates of bacterial sulfate reduction. Shallow lagoons could, according to [106], be favorable habitats for sulfur disporportionating bacteria before the rise of atmospheric oxygen. In addition, the mineral textures of pyrites studied here supports the assumption that the pyrite grains would have been formed in shallow water conditions in a zone away from the area where hydrothermal fluids vented onto the sea floor and where temperatures would have been low enough for the development of microbial activity.

6. Conclusions

A sulfur isotope study of sedimentary pyrite from Mesoarchean rocks of the Karelian Craton revealed signatures of mass-independent isotope fractionation and wide δ³⁴S variations in pyrite grains. Variations in the magnitude of ∆³³S values, in the range of −0.3‰ to +2.7‰, were close to those recorded in the rocks of similar ages in Southern Africa and Western Australia [22,23,24]. The negative and positive ∆³³S values observed in the present study were interpreted as a result of transferring photolytic sulfate and elemental sulfur to the sulfide sulfur in these samples; preservation of mass-independent signals in sedimentary rocks from the Karelian Craton supports the influence of atmospheric photochemistry on the sulfur cycling in the Mesoarchean [22].

The most striking features of the isotope data reported here are the remarkably large mass-dependent sulfur isotope fractionations where the δ³⁴S values extend from −10‰ to +32‰ and exceed previously reported δ³⁴S variations in Mesoarchean samples. In situ measurements of sulfide grains also reveal strong heterogeneity for both δ³⁴S and Δ³³S values. We argue that the source of these isotope effects was microbial processes, namely bacterial sulfur reduction in combination with microbial sulfur disproportionation. The present results can enhance understanding of sulfur cycling and microbial life in the Mesoarchean.