Atmospheric Influence of Earth's Earliest Sulfur Cycle

See allHide authors and affiliations

Science  04 Aug 2000:
Vol. 289, Issue 5480, pp. 756-758
DOI: 10.1126/science.289.5480.756


Mass-independent isotopic signatures for δ33S, δ34S, and δ36S from sulfide and sulfate in Precambrian rocks indicate that a change occurred in the sulfur cycle between 2090 and 2450 million years ago (Ma). Before 2450 Ma, the cycle was influenced by gas-phase atmospheric reactions. These atmospheric reactions also played a role in determining the oxidation state of sulfur, implying that atmospheric oxygen partial pressures were low and that the roles of oxidative weathering and of microbial oxidation and reduction of sulfur were minimal. Atmospheric fractionation processes should be considered in the use of sulfur isotopes to study the onset and consequences of microbial fractionation processes in Earth's early history.

The present-day sulfur cycle is strongly influenced by anthropogenic emissions, biological processes, and oxidative weathering of continental sulfides (1–3). It has been debated whether the sulfur cycle early in Earth's history was significantly different from the preanthropogenic sulfur cycle (4–9). Here we report sulfur multiple-isotope measurements (of δ33S, δ34S, and δ36S) of sulfide and sulfate minerals from Precambrian sedimentary and metasedimentary rocks and use them to document that a profound change occurred in the sulfur cycle between 2090 and 2450 Ma.

Thermodynamic, kinetic, and biological processes produce isotopic fractionations that depend on the relative mass differences between different isotopes of sulfur and oxygen. As a result, observed isotope variation can be related by δ33S = 0.515 × δ34S, δ36S = 1.90 × δ34S, and δ17O = 0.52 × δ18O (10). The quantities Δ33S, Δ36S, and Δ17O (11) reflect the deviation of measured isotope compositions (δ33S, δ34S, and δ36S, or δ17O and δ18O) from mass fractionation arrays with origins at δ33SCDT = 0, δ34SCDT = 0, and δ36SCDT = 0 for sulfur, and at δ17OSMOW = 0 and δ18OSMOW = 0 for oxygen (CDT, Canyon Diablo Troilite; SMOW, standard mean ocean water).

In addition, several isotopic fractionation processes are also known to produce mass-independent compositions (Δ33S ≠ 0, Δ36S ≠ 0, or Δ17O ≠ 0) (12–14). These include fractionations that result from hyperfine interactions in solid and liquid phases and an increasing number of gas-phase reactions (13). The hyperfine effect derives from spin-orbit coupling in isotopes with odd-mass nuclei and is therefore limited to isotopomers that contain these nuclei (such as 33S and 17O). Gas-phase mass-independent fractionations have been documented for a number of sulfur phases in laboratory experiments (SO2, H2S, CS2, and S2F10) (12, 15, 16), but the physical chemical origin of the effect is still uncertain. The role of gas-phase mass-independent chemistry in determining the oxygen isotopic compositions of many atmospheric species is unequivocal (13), and these types of reactions may also play an important role in determining the sulfur isotope compositions of atmospheric species (15).

To examine sulfur isotope variablility in early Earth, we extracted sulfur from sulfide and sulfate minerals in a variety of Precambrian samples. Sulfur isotopic analyses fall into two groups defined on the basis of Δ33S values and geological age (Fig. 1; also see Web table 1 Samples younger than 2090 Ma display a range of Δ33S values from –0.11 per mil (‰) to 0.02‰ and are considered to be consistent with fractionation by mass-dependent processes. Sulfide and sulfate samples older than 2090 Ma but younger than 2450 Ma exhibit a range of Δ33S varying between 0.02 and 0.34‰, and samples older than 2450 Ma exhibit a much larger range of Δ33S, varying between –1.29 and 2.04‰. This variation is consistent with large mass-independent compositions. The relation between Δ33S and Δ36S (Fig. 2) and between δ36S and δ34S for samples older than 3000 Ma (Fig. 3) rules out the possibility that hyperfine interactions account for the observations. The most likely explanation of observed Δ33S and Δ36S values for early Proterozoic and Archean sulfide and sulfate is, therefore, the presence of one or more gas-phase, mass-independent chemical reactions in the sulfur cycle.

Figure 1

Plot of Δ33S values versus sample age. Variable Δ33S values for samples with geological ages greater than the transition interval at 2090 to 2450 Ma are interpreted as an indication of atmospheric influence of the sulfur cycle. Homogenous Δ33S values for samples with geological ages less than 2090 Ma are interpreted as an indication of a sulfur cycle dominated by oxidative weathering of continental sulfide and sulfate. The dark gray band at Δ33S ∼ 0 represents the mean and 1 SD of data collected to date from younger sulfide and sulfate from 73 samples that include mantle xenoliths, marine barite, desert gypsum, evaporite, massive pyrite, channel deposits, ash deposits, and building surface deposits. Age data for these samples are compiled in (30).

Figure 2

Plot of Δ33S versus Δ36S for samples older than 2090 Ma. The correlation between Δ33S and Δ36S is interpreted as evidence of mass-independent isotopic fractionations originating in gas-phase reactions rather than from hyperfine interactions, which would produce Δ33S but not Δ36S. Mass-dependent processes plot at the origin of this plot. Error bars represent 1σ analytical uncertainties of 0.05 and 0.3‰ for Δ33S and Δ36S, respectively.

Figure 3

Three isotope plots of (A) δ36S versus δ34S and (B) δ33S versus δ34S for samples older than 3000 Ma. The array formed on (A) follows the relation δ36S = 2.17 (±0.1) × δ34S [calculated using York (31)] rather than the mass-dependent relation δ36S = 1.90 (±0.01) × δ34S (10). The data in (B) do not follow the tightly constrained mass-dependent relation δ33S = 0.515 (±0.005) × δ34S (10) but scatter on the diagram. Barite and chert-plus-barite samples are plotted as triangles. All other samples are plotted as diamonds. These arrays are inconsistent with biological fractionation processes and are consistent with a sulfur cycle that is strongly influenced by atmospheric chemical reactions and atmospheric oxidation reactions.

Photochemical reactions are thought to be important in the early Proterozoic and Archean atmosphere and may be relevant to the sulfur cycle before 2090 Ma. Photochemical reactions have been suggested as the source of nonzero Δ33S values in martian samples (martian meteorites) (15). Our data indicate that a profound change occurred in the sulfur cycle between 2090 and 2450 Ma. This change might represent the onset of a process capable of homogenizing mass-independently fractionated sulfur reservoirs or the suppression of one or more atmospheric reactions that had occurred before this interval.

Two basic models have been suggested for Earth's early sulfur cycle. The first is that the Archean sulfur cycle did not differ significantly from the preanthropogenic sulfur cycle (7–9) and that the dominant source of oceanic sulfate was oxidative weathering of continental sulfides and weathering of continental sulfates. The second is that oxidative weathering did not play a significant role in the Archean sulfur cycle and that the principal source of oceanic sulfate was photochemical oxidation of volcanogenic sulfur species in the Archean atmosphere (6). Our mass-independent sulfur isotope data strongly support a pre–2090-Ma sulfur cycle that was influenced by atmospheric chemical reactions. All of our data for samples of barite and of barite plus chert that are older than 2090 Ma have negative Δ33S values. One analysis of Archean barite from the Sargur Group of Karnataka, India (17), also yielded a negative Δ33S value. Most of our data for samples of pelitic and psammitic rocks older than 2090 Ma have positive Δ33S values. These observations are consistent with the existence of two reservoirs: a water-soluble (oceanic?) sulfate reservoir with negative Δ33S values and an insoluble reservoir of reduced sulfur with positive Δ33S values. Oxidative weathering would have mixed these reservoirs by transferring sulfur from reduced to oxidized reservoirs, producing sulfate with a positive (or juvenile) Δ33S signature that would dilute (or even eliminate) the negative Δ33S signature of its global oceanic counterpart. We infer that the transition to a sulfur cycle more like the modern preanthropogenic sulfur cycle occurred after 2090 Ma, when higher levels of atmospheric oxygen overwhelmed the atmospheric sources of oceanic sulfate through oxidative and microbial weathering of continental sulfides. Further insight into the nature of this transition will be obtained once the atmospheric reaction (or reactions) responsible for producing the effect are identified.

It is also possible that the atmospheric chemistry responsible for producing the observed mass-independent sulfur isotopic compositions before 2090 Ma may have stopped operating as a result of a change in atmospheric composition or actinic flux. Some have suggested that changes in the solar spectrum [ultraviolet (UV) and visible wavelengths] resulting from main sequence brightening (18) could have affected atmospheric chemistry in Earth's earliest atmosphere (19, 20). Although this change may account for our observations, changes in the abundance of absorbing species in the upper atmosphere exert a much stronger influence on lower atmospheric UV and on photochemistry (21). Higher surface UV resulting from a reduced ozone column depth in an atmosphere with low oxygen concentrations (22–24) may have played a role in determining atmospheric sulfur chemistry and in generating isotopically distinct sulfur reservoirs. Photolysis of SO2, for instance, results in production of SO3(25) that converts to H2SO4 upon contact with water. This photochemical oxidation sequence has been shown to produce mass-independently fractionated sulfur (15). Whereas oxygen isotopic signatures of present-day products of atmospheric oxidation of reduced sulfur-bearing gases have been shown to possess mass-independent oxygen isotopic compositions that are thought to be the result of oxidation by isotopically anomalous ozone and hydrogen peroxide (26,27), our oxygen isotope data for sulfate from barites from 3300 to 3400 Ma are mass-dependent (Δ17O = 0.01 ± 0.06‰). The lack of a similar mass-independent signature in our Archean barites may be a further indication that the sulfur oxidation pathways were different from that of the current atmosphere, possibly reflecting the exchange of sulfite (a product of SO2photooxidation) with water during the formation of sulfate. If this is the case, the buildup of atmospheric oxygen may have helped shut down the atmospheric chemical reactions that were responsible for generating isotopically anomalous sulfur-bearing reservoirs. An alternative explanation of the oxygen isotope data is that the original atmospheric oxygen isotopic signature has been lost because of exchange processes that occurred after sulfate formation.

If δ33S fractionations reflect atmospheric fractionation processes, it is also possible that the same applies to δ34S fractionations. Some workers have divided the sulfur isotope record into a period before 2750 Ma, when the range of δ34S was less than about 10‰, and a period after 2750 Ma, when the range of δ34S was greater than 10‰ (5, 28). These workers suggest that the change in magnitude for the range of observed δ34S fractionations at 2750 Ma is evidence of the onset of microbial sulfate reduction. Other workers have suggested that the smaller range of δ34S before 2750 Ma also results from microbial sulfate reduction but at higher temperature conditions, possibly at lower oceanic sulfate concentrations or in closed systems (7–9). On three isotope plots (Fig. 3), our data do not follow the mass-dependent arrays that would be formed by microbial sulfate reduction (for example, δ33S ∼ 0.515 × δ34S and δ36S = 1.90 × δ34S), and they allow that the δ34S record for samples older than 2750 Ma may be entirely atmospheric in origin. Although our data do not rule out the possibility of microbial sulfate reduction, they imply that δ34S values cannot be used alone to argue for the operation of (or examine the consequences of) metabolic processes that fractionated sulfur isotopes before 2750 Ma.

Similar sulfur isotope measurements can be used to resolve atmospheric inputs and to gain new insights into biogeochemical element cycles. The recent observation of mass-independently fractionated sulfur in this study and in martian meteorites (15) and of mass-independently fractionated oxygen in terrestrial and martian sulfates (27, 29) opens up possibilities for identifying additional components of the sulfur cycle on Earth and on other bodies in the solar system.


View Abstract

Stay Connected to Science

Navigate This Article