Explaining the Structure of the Archean Mass-Independent Sulfur Isotope Record

See allHide authors and affiliations

Science  09 Jul 2010:
Vol. 329, Issue 5988, pp. 204-207
DOI: 10.1126/science.1190298


Sulfur isotopes in ancient sediments provide a record of past environmental conditions. The long–time-scale variability and apparent asymmetry in the magnitude of minor sulfur isotope fractionation in Archean sediments remain unexplained. Using an integrated biogeochemical model of the Archean sulfur cycle, we find that the preservation of mass-independent sulfur is influenced by a variety of extra-atmospheric mechanisms, including biological activity and continental crust formation. Preservation of atmospherically produced mass-independent sulfur implies limited metabolic sulfur cycling before ~2500 million years ago; the asymmetry in the record indicates that bacterial sulfate reduction was geochemically unimportant at this time. Our results suggest that the large-scale structure of the record reflects variability in the oxidation state of volcanic sulfur volatiles.

Most natural processes fractionate sulfur in proportion to the mass difference between the isotopes (1). Ultraviolet (UV) photolysis of atmospheric SO2, however, produces a mass-independent fractionation (MIF) that is delivered to the surface only if atmospheric O2 levels are very low (27). The presence of MIF in sedimentary sulfides and sulfates older than 2450 million years (My) and its absence from later sediments has led to the accepted view that atmospheric O2 levels crossed a threshold value near the Archean-Proterozoic boundary (26). Beyond simply recording MIF, the Archean sulfur isotope record appears to carry a discernable temporal structure: moderate [<4 per mil (‰)] early Archean Δ33S (2) anomalies, followed by a mid-Archean minimum (<2‰) and a late Archean explosion in the magnitude of MIF (<12‰ in Δ33S). Previous studies attribute this variability to changes in the composition and oxidation state of the atmosphere and the associated evolution of photochemical pathways (79). In addition, an asymmetry in the record, with strongly positive but only weakly negative isotopic anomalies, remains without a quantitative explanation.

Here we explore the effect of a variety of extra-atmospheric processes on the characteristics and preservation of MIF. We present an integrated model of the full surface-sulfur cycle, accounting for the production and translation of atmospherically derived MIF through a marine reservoir and its preservation in the geologic record (10). We use recent measurements and theoretical calculations of 3xSO2 (x = 2, 3, 4, 6) UV absorption cross sections to constrain atmospheric MIF production (1012). By solving mass-balance equations for the steady-state reservoir sizes and isotopic compositions of four different oxidation states of sulfur [S6+ (sulfate), S4+ (sulfite, SO2), S0 (elemental sulfur), and S2– (sulfide, H2S)], we track MIF from production to lithification.

Rates of volcanic supply, photolytic destruction, gas-phase reactions, and net deposition to the surface govern the atmospheric lifetime of SO2. Any process that destroys atmospheric SO2 at the expense of photolysis reduces the production, by mass, of MIF [for example, atmospheric oxidation (Fig. 1A)], but as long as photolysis rates are non-negligible relative to the other atmospheric SO2 sinks, MIF is produced (though not necessarily preserved). In addition to nonphotolytic atmospheric sinks, which attenuate MIF by decreasing production, homogenization reduces MIF by remixing anomalous compositions back toward the original SO2 value. Whereas atmospheric oxidation to sulfate has been discussed in this context (6), microbial processes, which can perform a similar function, have not been rigorously investigated (Fig. 1B). Given a quantitatively important flux of MIF from SO2 photolysis, cycling between the sulfur reservoirs (e.g., microbial activity) or transformation to one oxidation state (e.g., quantitative reduction to sulfide) must be minor, as not to erase the anomaly. An immediate implication is that low atmospheric O2 is necessary but insufficient for preservation of MIF in the geologic record.

Fig. 1

The sensitivity of MIF in pyrite and sulfate, for Pco2 of 0.01 and 0.1 atm, plotted as a function of (A) the fraction of outgassed SO2 that is oxidized in the atmosphere, (B) microbial cycling, represented as a multiple of the sum of nonbiological sinks, (C) SO2:H2S in volcanic gases, (D) the disproportionation rate constant, (E) the hydrated formaldehyde concentration, and (F) the total sulfur (SO2 + H2S) outgassing rate. Metabolisms included in (C) are sulfate reduction, elemental sulfur oxidation, disproportionation and reduction, and sulfide oxidation. Estimates of modern biological sulfur cycling and volcanic outgassing are also shown.

Our model results illustrate the sensitivity of MIF to a few key properties of the ocean-atmosphere system, as well as its relative insensitivity to several other properties. Atmospheric deposition of SO2 leads to its speciation in seawater Embedded Image, where subsequent oxidation by Fe3+ (1315) leaves other aqueous oxidation pathways less important (for instance, Fe2+-catalyzed oxidation by aqueous O2 or by atmospherically produced H2O2). This leaves vanishingly little marine S4+ (~10−3 μM) and only modest sulfate concentrations (~102 μM). Given these oxidation rates, the absolute magnitude of MIF is only moderately sensitive to the adopted rate of S4+ disproportionation (Fig. 1D) (10), although the symmetry of the sulfate–elemental sulfur MIF is rate-sensitive. This is because the isotopic composition of SO2 propagates to both sulfate and elemental sulfur when disproportionation is rapid, but only to sulfate when oxidation dominates. The negative-positive asymmetry of the MIF record suggests that oxidation, not disproportionation, was the dominant aqueous S4+ sink. Hydrated formaldehyde complexes S4+, preventing its oxidation and disproportionation; however, this only affects MIF preservation at concentrations higher than those likely in an Archean ocean (Fig. 1E) (16).

The magnitude of MIF in pyrite, but not in sulfates, is very sensitive to the SO2:H2S ratio in volcanic volatiles (Fig. 1C). This is because the elemental sulfur that ends up in pyrite originates from both atmospheric SO2 and H2S, but H2S photoreactions do not generate MIF. When the outgassing rate of SO2 increases relative to that of H2S, more of the elemental sulfur budget comes from SO2 photolysis and, as a result, is anomalously fractionated. Sulfate, on the other hand, is produced almost entirely from oxidation and photolysis of SO2, and so its isotopic composition is insensitive to changes in the relative abundance of SO2 and H2S. Changes in the total sulfur (SO2 + H2S) outgassing rate with constant SO2:H2S produce no change in MIF magnitudes (Fig. 1F), but this may be due to the simplified nature of our atmospheric model; S8 production in more detailed atmospheric models is sensitive to the total sulfur outgassing rate (7, 17). We note, however, that more rapid S8 production does not necessarily translate into stronger MIF if the S8 is derived from H2S.

MIF depends critically on the partial pressure of CO2 (Pco2); increased Pco2 results in stronger scattering and UV absorption, decreasing SO2 photolysis rates (Fig. 2A). In more detailed atmospheric models, the relative abundances of atmospheric CO2 and CH4 also influence SO2 oxidation rates and the efficacy of MIF export (6, 7, 17). Our model highlights that, with a more acidic ocean (at high Pco2), the degree of pyrite saturation decreases, and H2S partitions more strongly into the atmosphere. As the importance of H2S photolysis relative to pyrite precipitation increases, more S0 is H2S-derived. This does not affect the Δ33S of sulfate (Δ33Ssulfate), but it decreases the Δ33S of pyrite [Δ33Spyr (Fig. 2B, dotted lines)] and influences the symmetry of the Δ33S signal.

Fig. 2

The sensitivity of (A) the SO2 column photolysis rate and (B) MIF in pyrite (black) and in sulfates (gray) to Pco2. Dashed lines represent the response of MIF to changes in atmospheric opacity due to increased molecular absorption and scattering by CO2, with Pco2 held constant at 0.01 atm in the ocean. Dotted lines represent the response to changes in the ocean (pH, Fe3+ concentration, etc.), with Pco2 held constant at 0.01 atm in the atmosphere. Solid lines show the combined effect.

Four properties emerge as important for the preserved magnitude and symmetry of Archean MIF: (i) atmospheric SO2 oxidation rates, (ii) redox transformations (including microbial cycling rates), (iii) SO2:H2S in volcanic gases, and (iv) Pco2. The very presence of MIF in the rock record points to low oxidant availability, suggesting that atmospheric oxidation was not the dominant SO2 sink (6, 7). With constraints from the geologic record on the remaining three properties, we explore their potential to explain the structure in the Archean MIF record using two quantities: Δ33Spyr and Rasym, a measure of the asymmetry Embedded Image. A successful explanation of the record must account for early Archean Δ33Spyr ≈ 4‰ and Rasym ≈ 2, mid-Archean Δ33Spyr ≈ 2‰ and Rasym ≈ 1, and latest Archean Δ33Spyr ≈ 11‰ and Rasym > 5.

When imposed on a purely abiological early Archean sulfur cycle, metabolic cycling between the different sulfur pools can potentially explain portions of the Archean MIF record (Fig. 1B). Both microbial S0 disproportionation and dissimilatory sulfate reduction may have existed since ~3500 million years ago (Ma) (18, 19). Adopting for the moment a scenario in which biological sulfur cycling is not important before ~3500 Ma, an increase in the role of microorganisms equivalent to ~10% modern cycling rates could drive a change in Δ33Spyr and Rasym similar to that observed from the early to mid-Archean (Fig. 3B, iii′). Difficulty arises, however, when microorganisms persist into the late Archean. Given that sulfate reduction delivers Δ33S < 0 to sulfide, a greater degree of symmetry (and even reversed asymmetry, Rasym < 1) would be expected for the late Archean; this is in marked contrast to the observed Rasym of 5 to 6 and suggests that dissimilatory sulfate reduction only rises to geochemical importance between 2400 and 2500 Ma, when high MIF magnitudes and large asymmetry are no longer observed. A late Archean or early Paleoproterozoic onset of sulfate reduction is consistent with a reanalysis of traditional sulfur isotope (δ34S) records (10) and suggests an explanation for the persistence of nonzero and relatively symmetric MIF for 10 to 100 My postdating the major loss of MIF at ~2500 Ma in some locations (5), but not in others (20).

Fig. 3

The dependence of preserved MIF on Pco2, volcanic SO2:H2S, and biological activity. (A) Color contours of pyrite Δ33S (Δ33Spyr) with no biological sulfur cycling, highlighting values relevant to the early, middle, and late Archean (solid black lines). Also shown are contours of the absolute value of the ratio of pyrite to sulfate Δ33S (Rasym, dotted white lines). (B) Same as in (A), but with biological cycling between sulfate, sulfur, and sulfide at ~10% modern values. (C) A section (iiiiii) through the phase space plotted in (A), with gradually evolving Pco2 (red line) and the value of volcanic SO2:H2S required to reproduce the observed Δ33Spyr and Rasym (thick gray line). The modeled Δ33S values of pyrite (thick solid black line) and sulfate minerals (thick dashed black line) are compared with the observed record (gray circles) in the lower half of (C). Regions iii′ in (A) and ii′ in (B) illustrate the effect of biological methanogenesis on Pco2 and the effect of microbial sulfur cycling on Δ33Spyr, respectively.

In the absence of biological sulfur cycling, we find that, with constant volcanic SO2:H2S, climatically reasonable changes in Pco2 (2123) cannot alone produce the observed MIF history (Fig. 3A). If we adopt instead an evolving value for Archean Pco2, calculated to offset changes in solar luminosity and to maintain liquid water [together with ~10 parts per million by volume atmospheric methane maintained by about twice modern sea-floor serpentinization rates (2124)], changes in volcanic SO2:H2S well within observed values (25, 26) easily account for the histories of both Δ33Spyr and Rasym (Fig. 3A, iiiiii). This result does not preclude further changes in Pco2 affecting the MIF signal [for example, through the effect of methanogens on Pco2 and Pch4 (21, 23)], though the asymmetry in the latest Archean almost by necessity indicates a substantial increase in volcanic SO2:H2S. Such an increase may be related to a major shift in the style of large igneous province eruption from submarine to subaerial in the late Archean (2700 to 2500 Ma), also suggested to have been the cause for the rise in atmospheric O2 and the loss of MIF (25). Consistent with the sulfur isotope record, our results suggest that this loss would be preceded by a MIF spike due to the elevated volcanic SO2:H2S. A small clustering of plume events during the early Archean [3500 to 3300 Ma (27)] may suggest an attendant increase in SO2:H2S, explaining the earliest MIF record, though it is not clear that these events were subaerial. The eruption rate required to raise global volcanic SO2:H2S from ~1.5 to ~8 (as in Fig. 3A, iiiii) is only about half to twice that of the Hawaiian plume (28), if the magma composition and SO2:H2S are comparable to those of Hawaii [1000 to 1500 parts per million sulfur, SO2:H2S of 50 to 100 (26)]. Given the plausibility of these changes, the structure of the Δ33S record may speak to large-scale crustal processes, superimposed onto and participating with low-O2 atmospheric chemistry.

Despite oceanic anoxia, rapid oxidation of SO2 by Fe3+ in the surface ocean suggests that, like the present, deposition into the Archean ocean was a terminal sink for atmospheric SO2 with a characteristic time scale of a few days. This implies that atmospheric SO2 was not well mixed and that spatio-temporal variability in its concentration is expected. Considering the demonstrated sensitivity of MIF to the relative abundance of SO2, this translates into the potential for large variability in pyrite Δ33S, with a much more muted signal in the Δ33S of sulfate minerals (Fig. 3). Additional variability in pyrite Δ33S is expected as a result of small-scale processes occurring in sediment pore water, given the possibility for sedimentary sulfides to adopt MIF from a variety of sources [dissolved sulfide, elemental sulfur, sulfite, or sulfate (17, 29)]. An aqueous oxidation origin for most of the oceanic sulfate may also explain the lack of correlation between Δ17O and Δ33S in early Archean barite (30).

Changes in atmospheric composition and chemistry probably explain a fraction of the variability in the Archean MIF record. In addition, we highlight the importance of variability in volcanic SO2:H2S, which may have caused the spike in MIF magnitudes ~2700 Ma, immediately preceding the ultimate loss of MIF brought on by the rise of atmospheric O2. In the absence of atmospheric O2 and a complex biosphere, it may have been the interactions between the solid Earth (mantle-crust) and fluid Earth (ocean-atmosphere) that drove the sedimentary record of Δ33S.

Supporting Online Material

SOM Text

Figs. S1 to S3

Tables S1 and S2


References and Notes

  1. MIF is expressed as the difference in per mil between measured isotopic compositions and those expected if the fractionations were mass-dependent: Δ33S ≡ δ33S – 0.515 × δ34S, Δ36S ≡ δ36S – 1.89 × δ34S.
  2. Further details are available in the supporting material on Science Online.
  3. Photooxidation of Fe2+ (14), thought to have been abundant in the Archean ocean (15), exceeds SO2 deposition under most circumstances (10).
  4. I.H. was supported by the Harvard University Origins of Life Initiative, and D.T.J. was supported by NASA Exobiology. D.P.S. thanks H. Breck and W. Breck for support. We also thank three anonymous reviewers.
View Abstract

Navigate This Article