Multiple sulfur-isotope signatures in Archean sulfates and their implications for the chemistry and dynamics of the early atmosphere

Significance

In an anoxic world, sulfate is rare or absent and therefore unlikely to be preserved in the geological record. It is puzzling, therefore, that several sulfate deposits were formed during the Archean. In situ S-isotopic fingerprinting of Archean barite (BaSO₄) revealed that three main sources of photochemical sulfur were involved in their formation. A strongly negative Δ³³S possibly generated at the point of volcanic emission, a slightly positive Δ³³S overlapping modern stratosphere-piercing volcanic sulfate aerosols, and a ³⁶S-anomalous sulfate source matching carbonyl sulfide photolytic products now preserved as background atmospheric sulfate aerosols. It is suggested that Archean barite deposits represent an exceptional period of volcanic sulfate aerosol production and preservation rather than an unexpectedly active biological sulfur cycle.

Abstract

Sulfur isotopic anomalies (∆³³S and ∆³⁶S) have been used to trace the redox evolution of the Precambrian atmosphere and to document the photochemistry and transport properties of the modern atmosphere. Recently, it was shown that modern sulfate aerosols formed in an oxidizing atmosphere can display important isotopic anomalies, thus questioning the significance of Archean sulfate deposits. Here, we performed in situ 4S-isotope measurements of 3.2- and 3.5-billion-year (Ga)-old sulfates. This in situ approach allows us to investigate the diversity of Archean sulfate texture and mineralogy with unprecedented resolution and from then on to deconvolute the ocean and atmosphere Archean sulfur cycle. A striking feature of our data is a bimodal distribution of δ³⁴S values at ∼+5‰ and +9‰, which is matched by modern sulfate aerosols. The peak at +5‰ represents barite of different ages and host-rock lithology showing a wide range of ∆³³S between −1.77‰ and +0.24‰. These barites are interpreted as primary volcanic emissions formed by SO₂ photochemical processes with variable contribution of carbonyl sulfide (OCS) shielding in an evolving volcanic plume. The δ³⁴S peak at +9‰ is associated with non–³³S-anomalous barites displaying negative ∆³⁶S values, which are best interpreted as volcanic sulfate aerosols formed from OCS photolysis. Our findings confirm the occurrence of a volcanic photochemical pathway specific to the early reduced atmosphere but identify variability within the Archean sulfate isotope record that suggests persistence throughout Earth history of photochemical reactions characteristic of the present-day stratosphere.

The amount of sulfate and its sulfur isotopic composition in the ocean through time is a function of the dynamic changes of sulfate sources (oxidative weathering on land, magmatic and hydrothermal input, and atmospheric photochemical reactions) and sulfate sinks (microbial and hydrothermal sulfate reduction and sulfate mineral precipitation). Although the Earth’s sulfate budget can be reasonably well constrained after ∼2.3 billion years (Ga) ago, when free oxygen became a permanent component of the atmosphere, our understanding of the ocean sulfate budget before 2.3 Ga ago is subject to uncertainties. The occurrence of mass-independent sulfur-isotope anomalies (MIF-S, noted Δ³³S and Δ³⁶S) in sedimentary sulfur (sulfide and sulfate) of Archean age (1) and the photochemical models (2) for the production and preservation of these anomalies support the view that the Archean atmospheric O₂ concentration was lower than 10⁻⁵ times the present atmospheric level. In this model, photochemical reactions involving volcanic SO₂ in the anoxic atmosphere yields both a reduced sulfur reservoir that can carry a highly positive Δ³³S and an oxidized sulfur reservoir with modestly negative Δ³³S, the Δ³⁶S values being of opposite sign. The corollary to this model is that sulfate influx from oxidative weathering on land should have been low before ∼2.3 Ga ago. This assumption is supported by mass balance model of the Archean sulfur cycle, which showed that the total weathering flux of sulfur to the ocean was negligible before 2.8 Ga and about three to five times lower between 2.8 and 2.5 Ga compared with the Proterozoic (3). Known Archean sulfate deposits occur as barite (BaSO₄) associated with felsic volcanic rocks in Western Australia, India, and South Africa at about 3.5, 3.4, and 3.2 Ga (4). How sulfate appeared in the oceans during the early Archean when oxidative weathering was absent remains unresolved. Does it reflect a period of unique conditions for the preservation of sulfate, an exceptional period of intense sulfate aerosol production, or an unexpectedly active biological sulfur cycle?

Barite is one of the best proxy for investigating S- and O-isotope processes during the Precambrian (5). In contrast to pyrite or sulfate mineral such as anhydrite (CaSO₄), barite is poorly soluble and less prone to dissolution/recrystallization processes, so that the S and O isotopes can hardly be reequilibrated over geological time scales. Hence, a key issue for elucidating the early Archean sulfur cycle concerns the significance of the isotopic composition of Archean barites. The main deposits in Australia and South Africa define a narrow range of both δ³⁴S values between +3‰ and +8‰ and Δ³³S anomalies between −0.1‰ and −1.2‰ (1, 5–11). In contrast to the δ³⁴S values that are relatively constant in all barite deposits, the Δ³³S values define a potential trend increasing from 3.5 (∼−1.5‰) to 3.2 Ga (∼−0.5‰; Fig. S1) (10). The significance of these isotopic variations is subject of debate (Fig. S2). It has been attributed to mixing between a photochemical sulfate, with both negative ∆³³S and δ³⁴S-values, representing the oxidized end product of a photochemical reaction that produced the so-called Archean reference array (ARA) of sulfides (possibly modified by bacterial processes), and an MDF sulfate (mass dependent fractionation, with ∆^33,36S = 0‰) pool of mantellic origin modified either by microbial sulfide oxidation (10) or by disproportionation of magmatic/hydrothermal SO₂ in the ocean (7). Thus far, the two sulfate end members leading to MIF and MDF signals have not been found in the geological record. This model also fails to explain why barite was deposited during restricted periods of time between 3.5 and 3.2 Ga (4, 12). Based on the analysis of microscopic pyrites in felsic volcanic rocks displaying Δ³³S/δ³⁴S relationships overlapping the isotopic composition of associated barite [Felsic volcanic array (FVA)], Philippot et al. (12) proposed that Archean barites could have formed temporally as volcanic ash deposits between 3.5 and 3.2 Ga ago (Fig. S2). In this model, most, if not all, sulfate may be of photochemical origin, with only minor overprint by MDF-sulfur processes. It also suggests that barite deposits should not be taken as representative of Archean oceans.

Fig. S1.

Plots of Δ³³S vs. δ³⁴S (A) and Δ³⁶S vs. Δ³³S (B) whole-rock Archean barite available in the literature. Blue crosses, Barite Valley (5, 10); open diamonds, Londozi (10); open circles, Vergelegen (10); open triangles, old Stentor (10); open square, Sargur (49); black crosses, Kangaroo Caves (9); red crosses, Dresser Formation (1, 6–8). The stars represent the average of the in situ analyses performed in this study on bedded barites of the Dresser (red) and Mapepe (blue) Formations and on disseminated barites of the Mapepe Formation present in felsic volcanic ash (gray) and other lithologies (ferruginous chert, volcanoclastic sediments, and spherule beds; black). Note the relationship between the age and ³³S-anomalies of barite.

Fig. S2.

Formation of sulfate–sulfide deposits in the Early Archaean. (A) Oceanic model in which sulfides defining the positive trend of slope ∼0.058 represents mixing between a mantellic reservoir (δ³⁴S = 0‰ and Δ³³S = 0‰) and sulfides formed by bacterial sulfate reduction (BSR) (7, 10). In this model, the oceanic sulfate reservoir originally lying on the ARA is transferred toward positive δ³⁴S values (between +3‰ and +8‰) by bacterial sulfate reduction. (B) Atmospheric model in which all sulfides and sulfates can be explained by isotope mass balance of S⁰ and sulfate aerosols of photolytic origin (FVA) (12). The dashed line represents a predicted photochemical array matching the sulfides trend of slope 0.058 ± 0.003 overlapping an unknown positive ∆³³S sulfate reservoir (positive MIF-SO₄?).

In situ analyses of Archean sedimentary sulfides with the secondary ion MS (SIMS) technique have revealed striking heterogeneities for δ³⁴S, Δ³³S, and Δ³⁶S at the grain and subgrain scales, which are not resolved by bulk SF₆ fluorination analyses (6, 12–14). Perhaps most importantly is that some in situ SIMS analyses yielded data in the four sulfur isotopes (δ³⁴S, Δ³³S, Δ³⁶S) that cannot be identified by bulk analyses (14). In contrast to sulfide, Archean sulfate has never been investigated for its multiple sulfur-isotope composition using an in situ technique. Here, we present results of SIMS analyses of the four sulfur isotopes in early Archean barite and explore these data with the specific goal of examining the origin and significance of the relationship between δ³⁴S, Δ³³S, and Δ³⁶S to address the different sulfur sources and photochemical processes involved in their formation.

Geological Setting

The studied barites are from the 3.49-Ga-old Dresser Formation, Warrawoona Group (Pilbara Craton, Western Australia) and the 3.26- to 3.24-Ga-old Mapepe Formation of the Fig Tree Group (Barberton Greenstone Belt, South Africa). The barite deposit of the Dresser Formation consists of a succession of chert, barite, volcano-clastic sediments, hydrothermal breccia, and carbonates deposited in an active volcanic caldera (15, 16). This succession, also referred to as chert-barite unit, is overlain by pillowed basalts and underlain by spinifex-textured metabasalts that experienced low-grade metamorphism between 100 °C and 350 °C (16). The underlying komatiitic basalts occur transected by an extended network of meter- to kilometer-scale barite and black and white chert veins indicating intense hydrothermal circulations (15, 16). The bedded chert-barite unit varies between 4 and 60 m thick. It is composed of predominantly bedded chert and thick units of coarsely crystalline barite in layers sometimes several meters thick and oriented parallel, or discordant, to the bedding (Fig. S3 A–C). The barite deposits of the Mapepe Formation overlie unconformably the deep-water carbonaceous black cherts and komatiitic volcanic units at the top of the Mendon Formation (Onverwacht Group). They consist of millimeter-scale to tens of centimeter-thick beds of barite, which are hosted by a complex clastic and cherty sequence including spherule beds, thin chert layers locally ferruginous, a variety of greenish carbonated cherts representing silicified and carbonated felsic ash and containing disseminated barite locally evolving into barite beds (17), cherty sandstones containing detrital barite and jasper grains, and chert-pebble conglomerates (18) (Fig. S3 D–J). These lithologies indicate the deposition in a complex, volcanically active, coastal and fan-delta association (17) controlled by the emergence and erosion of tectonically unstable highs leading to rapid changes of lithology (18). The rocks of the Mapepe Formation experienced relatively minor deformation and low-grade metamorphism under lower greenschist facies (19), which resulted in the good preservation of the original sedimentary and diagenetic structures. A network of black and white chert veins and the occurrence of local barite impregnation in the komatiitic rocks of the Mendon Formation indicate the circulation of hydrothermal fluids during or soon after deposition.

Fig. S3.

Geological setting and petrographic description of the different types of barite analyzed. (A) Massive type 1 bedded barite of the Dresser Formation attributed to a volcanic caldera (16, 17) (the scale is indicated by a person, arrowed). (B) Exposure of coarse-grained type 1 barite with low Δ³³S values from Dresser Formation. (C) Thin section optical view of a layer of microscale barite and pyrite laminates associated with coarse-grained barite of the Dresser Formation (sample 88.8 in PDP2b). (D and E) Field exposure (D) and corresponding section of core (E) intercepted by the drill hole BBDP of the bedded barite located at the base of the Mapepe sedimentary sequence (sample 78.18 in BBDP), showing the alternating nature of barite crystal fans and pyrite laminates. (F) Detail of sample 78.18 showing layers of microcrystalline barite (type 1 with intermediate Δ³³S values and type 2) interleaved with pyrite laminates. (G and H) Thin section optical view of a millimeter-scale barite layer composed of barite crystal oriented parallel to bedding within felsic volcanic ash and chert (type 1 barite with high Δ³³S values, sample 39.89 in BBDP). (I and J) Thin section optical view of a volcanoclastic sediment (sample 45.99 of BBDP) composed of quartz (Qtz) volcanic shards, carbonate (Carb), chlorite and mica (Phyl), barite (Ba, mixing of types 1 and 2), and pyrite (Py).

The samples investigated were collected from the Barberton Barite Drilling Project (BBDP) drill core, which shows the transition from the Mendon Formation (Onverwacht Group) to the shallow water terrigeneous and volcanoclastic units of the Mapepe Formation, and from the Pilbara Drilling Project (PDP2b and PDP2c) drill cores, which intercepted the chert-barite unit of the Dresser Formation (16, 20). A subset of samples studied by Philippot et al. (6, 12) for pyrite analysis has been selected for barite analysis with the aim of evaluating the heterogeneities in the four sulfur isotopes at the grain and subgrain scales. Two types of barites were analyzed: bedded barites collected either as surface sample (Pi-06-23) of the Dresser Formation or at different depths of the BBDP (39.89, 76.76, 77.38, 78.10, and 78.18) and PDP (88.7a, 89.3a, and 96.6a) drill cores. With the exception of one sample (39.89), which represents a millimeter-scale layer of barite interleaved with felsic volcanic ash and chert (Fig. S3 G and H), all bedded barite samples correspond to thick layers of crystalline barite. These layers are generally composed of two main types of barite including (i) coarse-grained, sometimes sector- and oscillatory-zoned, barite blades, locally forming crystal fans and (ii) randomly oriented microcrystalline barite (Fig. S3 B–F). Eight samples of nonbedded sedimentary barites have also been selected in volcano-clastic sediments (45.12, 45.99, 56.66, 76.27, and 76.36), spherule beds (68.00 and 68.05), and cherts (50.25) of the BBDP drill core. These barites consist of micrometer-scale, randomly disseminated grains of likely terrigeneous origin (Fig. S3 I and J).

Results

Results for analyses are presented in Table S1. Bedded barite from both Dresser and Mapepe Formations shows two main types of S-isotope compositions. Type 1 displays δ³⁴S values between +2.68 ± 0.31‰ and +5.44 ± 0.21‰ and ∆³³S values between −1.77 ± 0.26‰ and +0.24 ± 0.09‰ (Fig. 1A). This type can be further subdivided into three main subtypes on the basis of their geological provenance and ∆³³S values. The lowest ∆³³S values between −1.77 ± 0.26‰ and −1.15 ± 0.23‰ (∆³³S_mean ∼ −1.41‰) correspond to bedded barite from Dresser Formation. Barite with intermediate ∆³³S values between −1.26 ± 0.12‰ and −0.30 ± 0.09‰ (∆³³S_mean ∼−0.98‰) corresponds to meter-scale barite beds located at the base of the Mapepe Formation. The highest ∆³³S values between −0.40 ± 0.09‰ and +0.24 ± 0.09‰ (∆³³S_mean ∼−0.10‰) corresponds to a millimeter-scale barite layer of the Mapepe Formation interleaved within felsic volcanic ash and chert. Type 2 bedded barite shows higher δ³⁴S values between +7.67 ± 0.19‰ and +10.23 ± 0.25‰ and ∆³³S values between −0.90 ± 0.22‰ and −0.69 ± 0.22‰ for Dresser and −0.56 ± 0.10‰ and −0.02 ± 0.10‰ for Mapepe (Fig. 1A). This barite occurs in intimate association with type 1 bedded barite with low and intermediate ∆³³S values (Fig. S4). This relationship indicates that both types of barite deposited simultaneously during a same process. The preservation of grain-scale isotopic heterogeneities also suggests that the barite deposits represent primary precipitates, which were not subsequently rehomogenized in an open oceanic environment. Disseminated sedimentary barites present in terrigeneous volcano-clastic sediments, spherule beds, and laminated cherts of the Mapepe Formation show a range of δ³⁴S and ∆³³S values between +4.08 ± 0.28‰ and +11.00 ± 0.39‰, and −1.03 ± 0.09‰ and −0.18 ± 0.22‰, respectively (Fig. 1C). This range of composition is intermediate to the field defined by type 1 and type 2 bedded barites, which suggests that the disseminated barite have a mixed origin between these different sulfate sources. In a ∆³³S-∆³⁶S diagram, the different types of bedded barites show a different range of ∆³⁶S values (between +0.67 ± 0.51‰ and +2.12 ± 0.24‰ for type 1 with low ∆³³S, −1.37 ± 0.54‰ and +1.92 ± 0.29‰ for type 1 with intermediate ∆³³S, −0.91 ± 0.54‰ and +0.32 ± 0.54‰ for type 1 with high ∆³³S, and −1.58 ± 0.54‰ and +1.68 ± 0.42‰ for type 2; Fig. 1B). Taken individually, these different types define independent trends of steep slope centered on the ARA.

Table S1.

In situ sulfur-isotope analyses of Archean barites from Mapepe and Dresser Formations

Analysis	Lithology	δ34S	1σ	δ33S	1σ	δ36S	1σ	Δ33S	1σ	Δ36S	1σ
Mapepe Formation
39–89-bar1	Felsic volcanic ash	3.79	0.10	1.57	0.06	7.47	0.37	−0.38	0.09	0.32	0.54
39–89-bar2	Felsic volcanic ash	3.94	0.09	1.63	0.06	7.34	0.36	−0.40	0.09	−0.10	0.54
39–89-bar3	Felsic volcanic ash	3.81	0.11	1.85	0.07	7.39	0.39	−0.12	0.09	0.18	0.54
39–89-bar4	Felsic volcanic ash	3.94	0.10	1.93	0.07	6.76	0.37	−0.10	0.09	−0.69	0.54
39–89-bar5	Felsic volcanic ash	2.94	0.10	1.27	0.06	5.14	0.40	−0.24	0.09	−0.42	0.54
39–89-bar6	Felsic volcanic ash	4.30	0.09	2.11	0.07	7.52	0.37	−0.11	0.09	−0.61	0.54
39–89-bar7	Felsic volcanic ash	3.83	0.10	2.22	0.08	6.55	0.38	0.24	0.09	−0.69	0.54
39–89-bar8	Felsic volcanic ash	3.55	0.10	1.85	0.06	5.95	0.37	0.02	0.09	−0.76	0.54
39–89-bar9	Felsic volcanic ash	3.77	0.10	1.97	0.07	6.21	0.37	0.03	0.09	−0.91	0.54
39–89-bar10	Felsic volcanic ash	3.79	0.11	1.88	0.06	7.06	0.38	−0.08	0.09	−0.11	0.54
39–89-bar11	Felsic volcanic ash	4.09	0.10	2.17	0.07	7.56	0.39	0.07	0.09	−0.17	0.54
45–12-bar1	Volcano-clastic sediments	4.44	0.15	1.89	0.11	7.70	0.35	−0.40	0.09	−0.69	0.75
45–12-bar2	Volcano-clastic sediments	4.73	0.14	2.15	0.11	8.47	0.34	−0.29	0.09	−0.47	0.75
45–12-bar3	Volcano-clastic sediments	4.39	0.15	1.73	0.11	7.62	0.36	−0.53	0.09	−0.68	0.75
45–99-bar6	Volcano-clastic sediments	4.20	0.14	1.74	0.11			−0.42	0.09
45–99-bar7	Volcano-clastic sediments	7.14	0.14	2.65	0.11			−1.03	0.09
45–99-bar8	Volcano-clastic sediments	10.33	0.14	4.62	0.11			−0.70	0.09
45–99-bar9	Volcano-clastic sediments	4.08	0.14	1.65	0.11			−0.46	0.09
45–99-bar10	Volcano-clastic sediments	6.37	0.14	2.72	0.11			−0.56	0.09
50–25-bar2	Chert	6.02	0.12	2.60	0.14	11.15	0.35	−0.50	0.23	−0.24	0.51
50–25-bar3	Chert	5.92	0.12	2.64	0.14	10.80	0.36	−0.40	0.23	−0.38	0.51
50–25-bar4	Chert	6.53	0.12	3.11	0.14	11.24	0.34	−0.25	0.23	−1.10	0.51
50–25-bar5	Chert	7.43	0.12	3.45	0.14	14.00	0.34	−0.37	0.23	−0.05	0.51
50–25-bar6	Chert	4.89	0.12	1.86	0.14	9.58	0.34	−0.66	0.23	0.33	0.51
50–25-bar7	Chert	7.09	0.12	3.02	0.14	14.12	0.34	−0.63	0.23	0.72	0.51
50–25-bar8	Chert	4.04	0.12	1.80	0.14	7.67	0.35	−0.28	0.23	0.04	0.51
50–25-bar9	Chert	8.08	0.12	3.97	0.15	15.71	0.34	−0.20	0.23	0.43	0.51
50–25-bar10	Chert	8.39	0.12	3.84	0.15	16.84	0.36	−0.48	0.23	0.99	0.51
56–40-bar2	Volcano-clastic sediments	5.00	0.14	2.07	0.11	10.10	0.33	−0.51	0.09	0.65	0.75
56–40-bar3	Volcano-clastic sediments	5.76	0.14	2.07	0.11	11.82	0.33	−0.90	0.09	0.93	0.75
68–00-bar1	Spherule bed	4.97	0.20	1.85	0.15	9.45	0.51	−0.71	0.22	0.05	0.53
68–00-bar2	Spherule bed	4.67	0.20	1.64	0.15	9.61	0.50	−0.77	0.22	0.80	0.53
68–00-bar3	Spherule bed	6.43	0.20	2.48	0.15	11.85	0.51	−0.83	0.22	−0.31	0.53
68–00-bar4	Spherule bed	5.70	0.21	2.39	0.16	12.26	0.52	−0.55	0.22	1.49	0.53
68–05-bar1	Spherule bed	5.87	0.20	2.43	0.16	11.55	0.51	−0.59	0.22	0.46	0.53
68–05-bar2	Spherule bed	6.02	0.20	2.74	0.15	11.24	0.51	−0.36	0.22	−0.14	0.53
68–05-bar3	Spherule bed	6.26	0.20	2.93	0.15	11.47	0.50	−0.30	0.22	−0.36	0.53
68–05-bar4	Spherule bed	6.83	0.20	3.24	0.16	13.15	0.51	−0.28	0.22	0.23	0.53
76–27-bar1	Volcano-clastic sediments	9.31	0.12	4.58	0.15	17.79	0.36	−0.22	0.23	0.19	0.51
76–36-bar1	Volcano-clastic sediments	11.00	0.20	5.15	0.15	21.42	0.50	−0.52	0.22	0.62	0.53
76–36-bar2	Volcano-clastic sediments	10.91	0.20	5.44	0.15	21.18	0.51	−0.18	0.22	0.55	0.53
76–36-bar3	Volcano-clastic sediments	10.55	0.20	4.90	0.15	20.85	0.51	−0.53	0.22	0.90	0.53
76–76-bar1	Barite	3.98	0.11	0.81	0.11	8.86	0.29	−1.24	0.12	1.34	0.29
76–76-bar2	Barite	3.66	0.11	0.62	0.11	8.24	0.25	−1.26	0.12	1.32	0.29
76–76-bar3	Barite	3.71	0.11	0.79	0.11	7.36	0.28	−1.12	0.12	0.35	0.29
76–76-bar4	Barite	4.09	0.10	0.93	0.11	8.62	0.27	−1.18	0.12	0.88	0.29
77–38-bar1	Barite	3.41	0.10	0.70	0.11	7.73	0.24	−1.06	0.12	1.29	0.29
77–38-bar2	Barite	3.92	0.11	0.86	0.11	9.34	0.28	−1.16	0.12	1.92	0.29
77–38-bar3	Barite	3.61	0.11	0.78	0.11	7.95	0.27	−1.08	0.12	1.12	0.29
77–38-bar4	Barite	3.30	0.10	0.67	0.11	7.21	0.27	−1.03	0.12	0.97	0.29
77–38-bar5	Barite	5.31	0.10	1.67	0.11	11.32	0.24	−1.06	0.12	1.28	0.29
78–10-bar1	Barite	4.90	0.11	1.69	0.11	8.93	0.26	−0.83	0.12	−0.33	0.29
78–10-bar2	Barite	9.69	0.12	4.65	0.12	17.65	0.26	−0.34	0.12	−0.67	0.29
78–10-bar3	Barite	5.44	0.11	2.13	0.11	9.97	0.26	−0.67	0.12	−0.32	0.29
78–10-bar4	Barite	4.94	0.10	1.65	0.11	9.94	0.27	−0.89	0.12	0.61	0.29
78–10-bar5	Barite	3.99	0.10	0.88	0.11	8.42	0.25	−1.18	0.12	0.89	0.29
78–18-bar2	Barite	8.77	0.09	4.19	0.07	15.53	0.37	−0.33	0.09	−1.05	0.54
78–18-bar3	Barite	9.75	0.10	4.63	0.07	18.38	0.36	−0.39	0.09	−0.04	0.54
78–18-bar4	Barite	4.69	0.09	2.11	0.06	7.96	0.37	−0.30	0.09	−0.90	0.54
78–18-bar5	Barite	5.34	0.11	2.18	0.07	8.71	0.40	−0.57	0.09	−1.37	0.54
78–18-bar6	Barite	9.04	0.09	4.56	0.06	15.50	0.37	−0.10	0.09	−1.58	0.54
78–18-bar11	Barite	8.99	0.05	4.07	0.10			−0.56	0.10
78–18-bar12	Barite	9.41	0.08	4.44	0.11			−0.41	0.10
78–18-bar13	Barite	9.23	0.09	4.19	0.10			−0.57	0.10
78–18-bar14	Barite	8.69	0.10	4.04	0.11			−0.44	0.10
78–18-bar15	Barite	10.23	0.13	5.25	0.11			−0.02	0.10
Dresser Formation
Pi06-23-bar1	Barite	7.90	0.10	3.21	0.16	16.27	0.13	−0.85	0.22	1.34	0.42
Pi06-23-bar2	Barite	7.91	0.10	3.26	0.16	15.89	0.13	−0.82	0.22	0.94	0.42
Pi06-23-bar3	Barite	7.77	0.10	3.10	0.16	16.01	0.13	−0.90	0.22	1.32	0.42
Pi06-23-bar4	Barite	7.83	0.10	3.16	0.16	16.47	0.13	−0.87	0.22	1.68	0.42
Pi06-23-bar5	Barite	7.67	0.10	3.26	0.16	16.12	0.13	−0.69	0.22	1.61	0.42
88–7a-bar1	Barite	4.71	0.12	1.23	0.14	9.90	0.34	−1.20	0.23	1.00	0.51
88–7a-bar2	Barite	4.79	0.12	0.79	0.14	10.38	0.36	−1.68	0.23	1.33	0.51
88–7a-bar3	Barite	4.84	0.12	1.23	0.15	9.82	0.34	−1.27	0.23	0.67	0.51
88–7a-bar4	Barite	4.98	0.12	1.29	0.14	10.27	0.35	−1.28	0.23	0.85	0.51
88–7a-bar5	Barite	5.04	0.12	1.43	0.14	10.36	0.35	−1.17	0.23	0.83	0.51
88–7a-bar6	Barite	4.95	0.12	1.32	0.14	10.65	0.35	−1.23	0.23	1.29	0.51
88–7a-bar7	Barite	5.10	0.12	1.38	0.14	11.01	0.35	−1.25	0.23	1.37	0.51
88–7a-bar8	Barite	4.81	0.12	1.33	0.14	10.07	0.36	−1.15	0.23	0.97	0.51
88–7a-bar9	Barite	4.87	0.12	1.28	0.15	10.41	0.35	−1.23	0.23	1.20	0.51
89–3a-bar1	Barite	4.34	0.20	0.94	0.12	9.74	0.40	−1.29	0.17	1.55	0.42
89–3a-bar2	Barite	4.62	0.20	0.63	0.12	10.78	0.40	−1.75	0.17	2.04	0.42
89–3a-bar3	Barite	4.48	0.20	0.99	0.12	9.49	0.41	−1.32	0.17	1.02	0.42
89–3a-bar4	Barite	4.76	0.19	0.83	0.12	10.03	0.39	−1.62	0.17	1.03	0.42
89–3a-bar5	Barite	4.71	0.20	0.84	0.12	10.81	0.39	−1.58	0.17	1.90	0.42
89–3a-bar6	Barite	4.40	0.19	1.09	0.12	9.17	0.39	−1.18	0.17	0.84	0.42
89–3a-bar7	Barite	4.35	0.19	0.84	0.12	9.34	0.41	−1.40	0.17	1.12	0.42
89–3a-bar8	Barite	4.53	0.20	1.08	0.12	10.54	0.40	−1.26	0.17	1.98	0.42
89–3a-bar9	Barite	4.31	0.20	0.97	0.12	9.87	0.38	−1.25	0.17	1.72	0.42
96–6a-bar1	Barite	4.03	0.15	0.41	0.21	9.74	0.26	−1.67	0.26	2.12	0.24
96–6a-bar2	Barite	4.09	0.15	0.33	0.21	9.27	0.26	−1.77	0.26	1.54	0.24
96–6a-bar4	Barite	5.32	0.15	1.06	0.21	12.06	0.26	−1.68	0.26	2.01	0.24
96–6a-bar5	Barite	3.18	0.15	0.19	0.21	7.64	0.26	−1.45	0.26	1.62	0.24
96–6a-bar6	Barite	2.68	0.16	0.01	0.21	6.31	0.26	−1.37	0.26	1.25	0.24
96–6a-bar7	Barite	3.31	0.16	0.02	0.21	8.02	0.28	−1.68	0.26	1.76	0.24
96–6a-bar8	Barite	3.75	0.16	0.41	0.21	8.74	0.30	−1.52	0.26	1.66	0.24
96–6a-bar10	Barite	4.51	0.15	0.87	0.21	9.56	0.25	−1.46	0.26	1.04	0.24

Fig. 1.

In situ sulfur-isotope analyses of Archean sulfates. (A and B) Type 1 (circles) and type 2 (diamonds) bedded barites of the Dresser (open symbols) and Mapepe Formations (black symbols for meter-scale barite beds at the base of the Mapepe sedimentary sequence and red symbols for a millimeter-scale barite layer interleaved with felsic volcanic ash). (C and D) Disseminated barites of Mapepe Formation present in laminated chert, volcanoclastic sediment, and spherule bed (crosses). The 1σ error bars shown in Inset represent average uncertainties obtained on in situ sulfate analyses. The gray field represents the area defined by whole-rock analyses of Archean barite from the literature (1, 5–11, 49).

Fig. S4.

Example of ion probe barite analyses (red dots labeled 1–5) in sample 78.10 (BBDP) within a carbon-coated epoxy mount. Picture taking in reflected light mode (A). Plotted in a Δ³³S vs. δ³⁴S diagram (B), the data show the coexistence of type 1 (analyses 1, 3, 4, and 5) and type 2 (analysis 2) barite on a grain scale. Crosses represent disseminated barites of Mapepe Formation, whereas circles correspond to types 1 and 2 bedded barites of both Dresser and Mapepe Formations.

Discussion

A striking feature of our in situ barite analyses is a bimodal distribution of δ³⁴S values at about +5‰ and +9‰, which is matched by modern sulfate aerosols, but is not resolved by whole-barite analyses (single peak at +5‰; Fig. 2). This observation could be coincidental or indicates that Archean barite and modern sulfate aerosols have much in common (Fig. 3). Shaheen et al. (21) distinguished different sources of modern sulfate aerosols based on concentrations and δ³⁴S values of SO₄ in ice-core horizons. The low and intermediate δ³⁴S ranges (δ³⁴S = 1.4–2.6‰ and +12 ± 1.4‰) have been attributed to combining photoexcitation and photodissociation processes affecting different sulfur sources (mainly SO₂ and OCS) in the stratosphere. The most ³⁴S-enriched values correspond to seawater sulfate aerosols and to oxidation of organic compounds as DMS (δ³⁴S between +18‰ and +22‰). Accordingly, the two δ³⁴S peaks at ∼+5‰ and +9‰ (Fig. 2) suggest that similar SO_x photochemical processes were involved in the formation of Archean barite and modern stratospheric sulfate aerosols and that some of the isotopic differences (e.g., type 1 barite with different ∆³³S values; Figs. 1 and 3) could be explained by specific factors such as photolysis wavelength, gas pressure, and/or redox conditions.

Fig. 2.

Comparison of the frequency distribution of δ³⁴S composition in (A) present-day sulfate aerosols (n = 148, interval = 2‰) (21–24, 45, 50, 51) with (B) bulk (n = 140, interval = 1‰) 1, 5–11, 49), and (C) in situ (this study, n = 97, interval = 1.25‰) analyses of Archean barites. Modern and Archean sulfates define two main peaks at ∼+5‰ and +9‰ (highlighted in gray). The δ³⁴S of modern seawater sulfate (+21‰) and DMS (between +18‰ and +22‰) (52) are shown for comparison.

Fig. 3.

Comparison of (A) δ³⁴S-Δ³³S and (B) Δ³³S-Δ³⁶S compositions of Archean bedded and disseminated barites from Dresser and Mapepe Formations (open circles; this study) with modern sulfate aerosols (black circles) (21–24, 50). The 1σ error bars shown in inset represent average uncertainties obtained on in situ analyses of sulfate. Gray field labeled 190–220 nm represents the experimental results of SO₂ photolysis using UV of wavelength 190–220 nm (Xe lamp) (25, 26).

The δ³⁴S peak at ∼+5‰ is defined by the different type 1 bedded barite of Dresser and Mapepe Formations (Fig. 1). Type 1 bedded barite with high ∆³³S values (∆³³S_mean ∼−0.1‰ with a maximum value of +0.24 ± 0.09‰) is fully consistent with the range of modern stratospheric volcanic sulfate aerosols preserved in the ice and snow records (Fig. 3; δ³⁴S-∆³³S slope ∼0.07 ± 0.01 and ∆³³S-∆³⁶S slope ∼ −2.2 ± 0.4) (21–24), as well as laboratory experiments of SO₂ photodissociation in the 190- to 220-nm absorption region (δ³⁴S-∆³³S slope = 0.086 ± 0.035 and ∆³³S-∆³⁶S slope = −4.6 ± 1.3) (25, 26). This consistency between different sets of data, together with the observation that this type 1 barite is found in felsic volcanic ash, suggests that the same type of photolytic process, likely volcanic SO₂ photolysis, was effective both in the Archean atmosphere and modern stratosphere and potentially in the earliest Earth atmosphere (27). In contrast, type 1 barite with low and intermediate ∆³³S values is characteristic of Archean sulfate and was not identified in modern environments. Its origin is subject of debate. Previously identified by whole rock analyses, it has been attributed either to mixing between a photochemical sulfate with both negative ∆³³S and δ³⁴S values matching the ARA and a MDF seawater sulfate (7, 10) or to a photochemical sulfate reservoir of volcanic origin (FVA; Fig. S2) (12). Although our in situ analyses highlight the strong S-isotope heterogeneity of Archean barites (Fig. S4), no evidence of a photolytic sulfate pool matching the ARA could be identified in a δ³⁴S-∆³³S space. The good overlap between the FVA and the most negative type 1 bedded barite of the Dresser Formation provides therefore support for a volcanic photolytic origin of these sulfates. We argue below that the Dresser barite reflects specific photochemical processes occurring at the point of volcanic emission.

Studies of sulfate aerosols in nature and in the laboratory have generally focused on homogeneous chemical reactions in the stratosphere, but ignored the initial chemistry and history of the volcanic plume because primary volcanic aerosols emitted at volcanic centers cannot be mass independently fractionated on modern Earth (28). However, owing to the absence of oxygen shielding short wavelength solar UV (<350 nm), it is likely that MIF production occurred at the source of volcanic emissions during the Archean. In present day volcanoes, sulfur gases represent typically 2–35%/vol of volcanic gas emissions (29). The dominant sulfur component is sulfur dioxide (SO₂) followed by hydrogen sulfide (H₂S). The SO₂ fraction increases with decreasing pressure (30) and with increasing temperature and oxygen concentration of the magma (29, 31). Carbonyl sulfide (OCS) and its precursor carbon disulfide (CS₂) contribute a small fraction of 10⁻⁴% vol to 10⁻²% vol. In contrast to SO₂, OCS has a residence time of several years in the atmosphere, and modern volcanoes are considered to contribute less than 1% to the total global atmospheric OCS emission (32). In an anoxic world, the contribution of OCS to the MIF-S budget may have been much more significant than today. In reducing conditions, sulfur entering the atmosphere would be largely converted into OCS following the reaction 3CO + SO₂ = 2CO₂ + OCS (31). Ueno et al. (33) showed that high levels of carbonyl sulfide (OCS ∼ 1 ppm) would be expected if the plume is rich in CO (∼1%). In addition, previous studies of early planetary evolution predicted that OCS could be an important primary S volcanic species (34). Using a one-box model and considering a volcanic sulfur flux >3× larger than the modern volcanic outgassing rate (35), Ueno et al. (33) suggested that about 5 ppm of OCS is needed to explain the Archean Δ³³S record. The OCS molecule presents a large absorption cross section between 190 and 260 nm (36) and therefore has the ability to prevent SO₂ photolysis at UV wavelength greater than 202 nm. SO₂ photolysis between 180 and 200 nm produces sulfate with negative ∆³³S (33), which is typical of Archean barite. Accordingly, it is suggested that the sulfate source involved in the formation of type 1 barite could represent primary volcanic emissions formed by SO₂ photolysis with variable contribution of OCS shielding in an evolving volcanic plume (Fig. 4).

Fig. 4.

Schematic model for the formation and deposition of Archean sulfate aerosols. (A) Different stages of volcanic activity and associated environments of barite deposition. (B and C) Sulfur isotopic compositions of barite (open circle; this study) (1, 5–11, 49) and sulfides (black square) (7–9, 49, 53) from Archean barite deposits and associated volcano-clastic and terrigenous sediments. Most of the sulfide isotopic compositions define a trend of slope 0.058 ± 0.003 (95% CI), similar to the experimental results of SO₂ photolysis using UV of wavelength of 190–220 nm (Xe lamp; red field) (25, 26). The green area represents the experimental results of OCS photolysis at λ < 200 nm (Xe lamp) (44). FVA (blue trend) is interpreted as OCS-shielded photolysis of SO₂ at λ < 202 nm.

The Δ³³S variations recorded for Type 1 bedded barite of different ages (3.5 Ga old barite with low Δ³³S values, and 3.2 Ga old barite with intermediate and high Δ³³S values; Fig. S1) indicate that the Δ³³S of the photochemical sulfate products was not constant both in time and space. The most negative Δ³³S values associated with the massive bedded barite and vein network of the Dresser Formation could represent primary sulfate deposits formed at the point of volcanic emission where the density of SO₂ can increase by orders of magnitude in the days to weeks following a large eruption. This interpretation is supported by geological relationships indicating that Dresser barite formed in an active volcanic caldera (15, 16). Part of this SO₂ will react with CO to form OCS. OCS shielding combined with SO₂ shielding would in turn generate large amounts of sulfate and elemental sulfur aerosols with specific S-isotope characteristics, which would quantitatively overprint the isotope signal emerging from the background atmosphere. None of the broadband or single line SO₂ photochemical experiments available today can be used to account for the photochemical reactions occurring in dense volcanic plumes. Although displaying a negative slope subparallel to the FVA trend, SO₂ photolysis results using 193 nm UV (37) cannot be considered because of the extreme sensitivity of SO₂ isotopologue absorption cross-sections to UV wavelength (38). Accordingly, SO₂ photolysis experiments performed under various P-T-X conditions and integrating over the 180- to 202-nm UV range are needed to evaluate the photochemical significance of the FVA. In contrast, as discussed above, the high Δ³³S values (Δ³³S_mean ∼ −0.1‰) mimicking modern stratospheric sulfate aerosols may have formed at some distance from the point of emission through interaction with chemically and optically homogeneous volcanic plumes containing minor amounts of OCS. Intermediate ³³S-anomalous type 1 bedded barite of the Mapepe Formation (∆³³S_mean ∼ −1‰) could represent mixing between these two end member sulfate sources.

The δ³⁴S peak at ∼9‰ is defined by type 2 bedded barite of the Dresser and Mapepe Formations (∆³³S up to 0‰; Fig. 1). The near-zero ∆³³S values could indicate a sulfate source of nonphotochemical origin, such as the mantle, and the occurrence of positive δ³⁴S may highlight a MDF process, such as disproportionation of volcanic SO₂ (7) or microbial sulfate reduction (10). Hence, this ³⁴S-enriched sulfate reservoir may be representative of Archean seawater (7, 10). However, the negative ∆³⁶S values (−1.6 ± 0.5‰) recorded by the type 2 barite do not support these interpretations (Fig. 1) (39, 40). Shaheen et al. (21) reported anomalous sulfate aerosols with similar negative Δ³⁶S (−0.6 ± 0.2‰) formed either through OCS photolysis in the stratosphere or nonphotochemical biomass and fossil fuel burning in the troposphere. Biomass and fossil fuel burning are unlikely for Archean environments. Hence, by a process of elimination, OCS photolysis remains the only process available to account for type 2 barite. On modern Earth, OCS is stored in the near surface of oceans and is released to the atmosphere through hydrolysis in seawater (41). The low solubility and long atmospheric lifetime with respect to tropospheric chemistry and photolysis enable a significant fraction of OCS to reach the stratosphere (42), where it photo-dissociates to carbon monoxide and elemental sulfur (S⁰) on UV irradiation in the 200- to 260-nm range. The sulfur atom is oxidized by different pathways (OH/H₂O₂/CO₂/O₃/O₂) to SO₂ and ultimately sulfate, which is thought to contribute to the stratospheric sulfate aerosol layer (21, 43). Laboratory experiments have shown that OCS photolysis at short wavelengths does not produce ³³S-isotope anomaly in oxygen-free conditions but negative ∆³⁶S down to −1.5‰ (Fig. S5) (44). Accordingly, considering that OCS may have been an important sulfur component of the volcanic plume, it is suggested that sulfate aerosols could have formed through photolysis of OCS and oxidation of elemental sulfur by OH and/or H₂O₂ (45), the latter being efficiently generated in a CO₂/CH₄-rich and O₂-depleted Archean atmosphere (46). Direct photolysis and oxidation of OCS gases to sulfate aerosols on mineral and dust surfaces within the volcanic plume (47) could explain that both type 1 and type 2 barites coexist in a same sample (Fig. S4).

Fig. S5.

Comparison of the sulfur isotopic compositions of product S⁰ (black squares) obtained by OCS (open star) photolysis experiments at >200 nm (Xe lamp) (44) with type 2 barite from the Mapepe Formation (open circles; this study).

Conclusions

The presence of S-MIF in the present-day atmosphere in nonvolcanic aerosols after the super ENSO 1997–1998 event has led Shaheen et al. (21) to suggest that both SO₂ and OCS sources could have contributed to sulfur-isotopic anomalies in the Archean. Our results fully support this hypothesis. In our model, however, considering the reducing nature of the early Earth, both SO₂ and OCS photolysis leading to elemental sulfur and sulfate aerosol production can occur during periods of active volcanism (Fig. 4). The origin of the strongly depleted ³³S-anomalies in Archean barites remains enigmatic. Our results tend to support the view of a photochemical process taking place near the point of volcanic emission. The recent findings by Roerdink et al. (48) that the different Archean barite deposits display different Sr-isotope ratios typical of the continental crust that are correlated with both ³³S-anomalies and barite age provide further support for a local (volcanic) rather than a global (marine) origin of Archean sulfate. If correct, this implies that Archean barites can be considered as important targets for studying the dynamics and atmosphere chemistry of early Earth but not earliest environments and emergence of microbial metabolisms. Because S isotopes of coexisting sulfate and sulfide involved in bacterial sulfate reduction and photochemical processes should follow specific trends, the feasibility of this claim could be tested in future studies by careful evaluation of sulfate and sulfide formation pathways and the existing geologic record.

Materials and Methods

In situ analyses of barite were performed with the CAMECA IMS 1280HR ion microprobe at Centre de Recherches Pétrographiques et Géochimiques (CRPG) with previously described procedures (12). The ion probe settings are basically the same for both the analysis of sulfides and sulfates (Table S2). Details of the analytical settings are described in SI Materials and Methods and in Fig. S6. The uncertainties (1σ) on δ³⁴S, ∆³³S, and ∆³⁶S are, respectively, around ±0.2‰, ±0.2‰, and ±0.5‰ on average.

Table S2.

Description of the analytical settings for multiple sulfur measurements in pyrites and in barites

Analytical settings	Pyrite	Barite
Instrument	CAMECA IMS 1280HR
Primary ions	Cs+
Primary beam intensity	∼5 nA	∼10 nA
Beam diameter	∼15 µm	∼20 µm
Primary/secondary voltages	+10 kV/−10 kV
Electron gun	Yes
Presputtering time	300 s	60 s
Raster	10 µm
Transfer field magnification	90×
Field aperture	2,500 × 2,500 µm
Entrance slit	60 µm
Multicollection slit, n	n = 2
Contrast aperture	400 µm
Energy window	50 eV with an offset of 5 eV
Detectors	L'2(FC), C(FC), H1(FC), H2(EM)
Resistors of the FC amplifiers	1010 Ω for L'2, and 1011 Ω for C and H1
FC background measurement	During the presputtering
Number of cycles	30	30–40
Counting time	4 s for each cycle
34S− count rate	>109 cps	>3.108 cps
Magnetic field control	NMR regulation

cps, counts per second; EM, electron multiplier; FC, Faraday cup.

Fig. S6.

Within-run drift of δ³⁴S measurement relative to the first cycle analysis in barite.

SI Materials and Methods

In situ analyses of barite were performed with the CAMECA IMS 1280HR ion microprobe at CRPG with previously described procedures (12, 54). All isotopic compositions are reported as per mil (‰) deviations from V-CDT standard using the conventional notation

$$\mathrm{\delta }{}^{34}\text{S}=\left[{({}^{34}\text{S}/{}^{32}\text{S})}_{\text{sample}}/{({}^{34}\text{S}/{}^{32}\text{S})}_{\text{V}-\text{CDT}}-1\right]\times \mathrm{1,000}.$$

∆³³S and ∆³⁶S are used as a measure of the deviation from the terrestrial mass-dependent fractionation line and represent the MIF expressed as (1, 39)

$$\mathrm{\Delta }{}^{33}\text{S}=\mathrm{\delta }{}^{33}\text{S}-\mathrm{\delta }{}^{34}\text{S}\times 0.515,$$

$$\mathrm{\Delta }{}^{36}\text{S}=\mathrm{\delta }{}^{36}\text{S}-\mathrm{\delta }{}^{34}\text{S}\times 1.89.$$

The ion probe settings are basically the same for both the analysis of sulfides and sulfates. Briefly, mounts of polished thin sections of the samples with selected sulfates were sputtered with a 20-µm Cs⁺ primary ion beam of 10 nA. Secondary, ^32,33,34S⁻ ions were analyzed in multicollection mode using three off-axis Faraday cups (L′2, C, and H1) that were intercalibrated before each session of analyses, and ³⁶S⁻ ions were measured on an electron multiplier (H2). Typically, ³²S⁻ count rates are >3.10⁸ cps for barite (Table S2). The normal incidence electron gun was used for charge compensation. The drift of the EM due to its aging was corrected using a routine in the Cameca CIPS software (more details in ref. 54). The measurements were performed using slit n°2 of the multicollection and an entrance slit width of ∼60 µm. These analytical settings allow obtaining a mass resolving power (M/∆M) of roughly 5,000. Under these conditions, the isobaric interferences for sulfur isotopes, particularly due to hydrides XH, are fully resolved. Measurements were performed with an automatic centering process and consist of 30 cycles of 60 s (Table S2). Internal error is less than 0.05 ‰ for δ³⁴S, δ³³S, and δ³⁶S.

The stability of the instrument was monitored by repeated measurements of in-house barite reference material Morvan (average bulk values, δ³⁴S = +19.26 ± 0.18‰, ∆³³S = 0.04 ± 0.02‰, and ∆³⁶S = 0.32 ± 0.20‰, 2σ, n = 4). Each series of 10–20 sample analyses have been bracketed by three to six analyses of Morvan. The instrumental fractionation ε is defined from the average of the raw ratios of SIMS measurements and the δ values obtained from conventional method following

$${\mathrm{\epsilon }}^{3\text{x}}={\left({\mathrm{\delta }}^{3\text{x}}{\text{S}}_{\text{raw}}\right)}_{\text{average}}-{\mathrm{\delta }}^{3\text{x}}{\text{S}}_{\text{true}}\left(\mathrm{for}\hspace{0.17em}{\mathrm{\epsilon }}^{34},{\mathrm{\epsilon }}^{33},{\mathrm{\epsilon }}^{36}\right).$$

The within-run drifts (Fig. S6) have been time corrected for each bracketed series with a linear regression

$${\mathrm{\epsilon }}_{\text{TC}}^{3\text{x}}={\mathrm{\epsilon }}^{3\text{x}}+\text{mx},$$

where m is the slope of the regression and x the analysis number between bracketing standards. Then, raw data have been corrected for instrumental fractionation following

$${\mathrm{\delta }}^{3\text{x}}{\text{S}}_{\text{TC}}={\mathrm{\delta }}^{3\text{x}}{\text{S}}_{\text{raw}}-{\mathrm{\epsilon }}_{\text{TC}}^{3\text{x}}.$$

The mass independent fractionations ∆³³S and ∆³⁶S are then calculated with the corrected δ values. The errors on δ values take into account the internal error and the reproducibility of the standards, whereas those on ∆³³S and ∆³⁶S correspond only to the reproducibility of the standards bracketing each series. The uncertainties (1σ) on δ³⁴S, ∆³³S, and ∆³⁶S are, respectively, around ±0.2‰, ±0.2‰, and ±0.5‰ on average.

Instrumental Settings and Accuracy of in Situ S-Isotope Analysis.

A detailed description of the analyses of the four sulfur isotopes using the Cameca 1280 can be found in ref. 54. The details of the analytical protocol used for sulfide vs. sulfate analysis is summarized in Table S2. Most instrumental settings (mass resolution, detectors, field aperture) used for barite S-isotope analysis are similar to the one used for pyrite analysis. The parameters that differ include (i) the intensity of the primary beam (5 nA for pyrite vs. 10 nA for barite) because of the different emission yield between these minerals, and (ii) the time of presputtering (300 s for pyrite vs. 60 s for barite). Fig. S6 shows that the within-run evolution of barite is limited to ±3‰ over 40 cycles, which is in good agreement with previous studies (see ref. 54 for a discussion).

The frequency of δ³⁴S values obtained by bulk and in situ techniques (Fig. 2) are remarkably overlapping, thus indicating that the two techniques yielded the same results. Taken individually, the data reveal that barite can be isotopically heterogeneous on a grain scale. This observation implies that for each sample a large number of analyses would be required to match each bulk analysis, which is unrealistic both in terms of machine access and costs. To demonstrate the veracity of our results, we compared the bulk and in situ S-isotope data of sample Pi06-23, which is composed of large and relatively homogenous barite crystals. The bulk isotopic composition of sample Pi06-23 is δ³⁴S = +6.96 ± 0.10‰, ∆³³S = −0.88 ± 0.01‰, and ∆³⁶S = 1.09 ± 0.08‰ (1σ, n = 2). These values are similar within errors of the mean values of our in situ data δ³⁴S = +7.82 ± 0.10‰, ∆³³S = −0.83 ± 0.08‰, and ∆³⁶S = 1.38 ± 0.29‰ (1σ, ν = 5 analyses shown in Table S1).