The Rise of Atmospheric O2: New Insights From Mass-Independently Fractionated Sulfur Isotopes James Kasting Department of Geosciences Penn State University

Talk outline Part 1—Background on the rise of atmospheric O2 Part 2—Sulfur isotopes and the Archean MIF (mass-independent fractionation) record Part 3—Causes of sulfur MIF: The chain formation mechanism

Geologic O2 Indicators Blue boxes indicate low O2 (Detrital) H. D. Holland (1994) Blue boxes indicate low O2 Red boxes indicate high O2 Dates have been revised; the initial rise of O2 is now placed at 2.45 Ga

Geologic O2 Indicators (Detrital) H. D. Holland (1994) The marked increase in atmospheric O2 inferred at 2.2 (or 2.45) Ga is often referred to as the Great Oxidation Event, or GOE

Caprock Canyon (Permian and Triassic) redbeds Redbeds contain oxidized, or ferric iron (Fe+3) Fe2O3 (Hematite) Their formation requires the presence of atmospheric O2 Reduced, or ferrous iron, (Fe+2) is found in sandstones older than ~2.2 b.y. of age http://www.utpb.edu/ceed/GeologicalResources/West_Texas_Geology/links/permo_triassiac.htm

Witwatersrand gold ore with detrital pyrite (~3.0 Ga) • Pyrite = FeS2 • oxidized during weathering today  Atmospheric O2 was low when this deposit formed P. Cloud, Oasis in Space (1988)

Banded iron- formation or BIF (>1.8 b.y ago) • Fe+2 is soluble, while Fe+3 is not • BIFs require long- range transport of iron  The deep ocean was anoxic when BIFs formed

What caused the rise of O2?

What caused the rise of O2? We argue about exactly why O2 rose at ~2.5 Ga, but we all agree that the initial rise in O2 was caused by cyanobacteria These are the only prokaryotes that are capable of oxygenic photosynthesis Higher plants and eukaryotic algae clearly obtained this capability from them

Different forms of cyanobacteria (formerly “blue-green algae”) a) Chroococcus b) Oscillatoria c) Nostoc (coccoid) (filamentous) (heterocystic) Nitrogen-fixing

Trichodesmium bloom Fix N in the morning Produce O2 in the afternoon Berman-Frank et al., Science (2001)

A big question is whether oxygenic photosynthesis was Science (2007) A big question is whether oxygenic photosynthesis was invented at 2.45 Ga or whether it was invented somewhat earlier, and so something else triggered the GOE Evidence from Mb and Re suggests that O2 was being produced before the GOE

Back to the sulfur isotope story… Despite the impressive geologic evidence for a GOE near 2.5 Ga amassed and interpreted by Preston Cloud, Dick Holland, and others, skeptics still remained This all changed, though, in 2000 as a consequence of studies of sulfur isotopes…

Part 2—Sulfur isotopes and the Archean MIF (mass-independent fractionation) record

S isotopes and the rise of O2 Sulfur has 4 stable isotopes: 32S (95%), 33S (0.75%), 34S (4.25%), and 36S (0.01%) We measure their abundances relative to an isotopic standard, typically CDT (Canyon Diablo troilite, FeS)…

S isotopes and the rise of O2 In thermodynamic equilibrium, the isotopes follow a mass-dependent curve with xS = 0 with Here, (x) = 0.516 for 33S = 1.89 for 36S

Mass-dependent isotope fractionation Vibrational energy levels depend inversely on the reduced mass  = (k/mR)1/2 En = (n+½) h Increasing the mass of one or both atoms decreases the vibrational frequency and energy, thereby strengthening chemical bonds A simple harmonic oscillator

S isotopes and the rise of O2 For small values of , one can linearize this equation and define a standard mass fractionation line (MFL) 33S  0.516 34S 36S  1.89 34S But, in very old (Archean) sediments, the isotopes fall off this line 

S isotopes in Archean sediments Farquhar et al. (2001) (FeS2) (BaSO4) 33S In these initial data, sulfides (pyrite) fall above the mass fractionation line; sulfates (barite) fall below it This phenomenon is often termed ‘mass-independent fractionation’, or MIF

The paper that broke the sulfur MIF story was published in 2000 by James Farquhar, Huming Bao, and Mark Thiemens 

33S versus time High O2 Low O2 Farquhar et al., Science, 2000 73 Phanerozoic samples Farquhar et al., Science, 2000

MIF data as of 2011 The pattern in 33S continues to hold as new data are added Note the significant asymmetry between positive and negative 33S values, especially from 2.5-2.7 Ga Need a larger sulfur exit channel for the negative values so that the isotopic signal is diluted D. T. Johnston, Earth Sci. Rev (2011)

Additional patterns in the S isotope data 33S is positively correlated with 34S (slope ~1), especially between 2.5 and 2.7 Ga Here, 34S is the fractional deviation from the sulfur isotope standard, typically CDT (Canyon Diablo Troilite, FeS) D. T. Johnston (2011)

Additional patterns in the S isotope data In recent years, geochemists have been able to measure the the much less abundant isotope 36S 36S is negatively correlated with 33S (slope ~ 0.9) From Babikov, PNAS, in press

Part 3—Causes of sulfur MIF: The chain formation mechanism

Causes of sulfur MIF All of this has led to great interest in the causes of sulfur MIF Ever since 2000, the leading candidate has been photolysis of SO2 in the 190-220 nm band SO2 + h  SO + O The SO2 molecule predissociates (i.e., goes to an excited bound state before breaking apart), creating lots of opportunities for unusual isotope effects

Production of sulfur “MIF” by SO2 photolysis excitation UV absorption coefficients Blowup of different forms of SO2 The different isotopologues of SO2 (e.g. 32SO2 and 33SO2) absorb UV radiation at slightly different wavelengths J.R. Lyons, GRL (2007)

Production of sulfur “MIF” by SO2 photolysis excitation UV absorption coefficients Blowup of different forms of SO2 This wavelength region is shielded by ozone today, and so Farquhar et al. inferred that the Archean atmosphere must have been low in O2 and O3 J.R. Lyons, GRL (2007)

Farquhar et al. were basically correct, but there is another way to think about this that is in some ways more general and that leads to a firm upper limit on atmospheric O2 during the Archean…

Astrobiology (2002) Pavlov and Kasting modeled atmospheric sulfur chemistry using a 1-D photochemical model 100 layers up to 100 km 72 chemical species (plus sulfur isotopic species) 337 photochemical reactions The MIF signal was assumed to have been created by SO2 photolysis, and nothing else

Archean sulfur cycle In a low-O2 atmosphere, volcanic SO2 can be either oxidized or reduced (or it can exit the atmosphere as SO2) By contrast, today, virtually all SO2 is oxidized to sulfate; thus, any MIF signal is eliminated by homogenization Kasting, Science (2001) [Redrawn from Kasting et al., OLEB (1989)]

“Archean” sulfur removal rates Typically, SO2, H2S, sulfate, and S8 are the main exit routes for sulfur from a low-O2 atmosphere Pavlov and Kasting (2002)

Result (still valid today): pO2 during the Archean must have been <10-5 PAL (times the Present Atmospheric Level)

Result (still valid today): pO2 during the Archean must have been <10-5 PAL (times the Present Atmospheric Level) But, Pavlov and Kasting appear to have made a significant mistake in doing their calculation…

Pavlov and Kasting assumptions In order to generate a system of chemical reaction that did not generate MIF, Pavlov and Kasting adopted the following rules:

PNAS, in press In a new paper, Babikov has shown that while assumptions 2 and 3 are correct, assumption 1 is valid only some of the time In particular, isotopic reaction rates should be doubled for S2 formation, but not for formation of S4 or S8

Sulfur chain formation The main formation sequence is thought to be S + S + M  S2 + M S2 + S2 + M  S4 + M S4 + S4 + M  S8 + M At each of the latter two steps, the chance of incorporating any minor isotope, xS, goes down by a factor of 2  big MIF signal that affects each isotope equally S2 S4 S8

Sulfur chain formation S2 formation does not create MIF because S2 acts as rigid rotator The ‘normal’ (32S32S) molecule is symmetric, so it lacks even rotational states The isotopically substituted molecule (xSS) is asymmetric and has both even and odd rotational states This creates twice as many pathways for formation of xSS compared to S2, thereby doubling the rate constant S2 S4 S8

Sulfur chain formation S4 and S8 are also symmetric molecules, but they are ‘floppy’ and should not behave like rigid rotators; hence, they do not obey the same selection rules on rotational states, and so the rates of the isotopically substituted reactions should not be doubled S2 S4 S8

Sulfur chain formation S4 and S8 are also symmetric molecules, but they are ‘floppy’ and should not behave like rigid rotators; hence, they do not obey the same selection rules on rotational states, and so the rates of the isotopically substituted reactions should not be doubled S2 S4 S8

MIF generation during formation of S4 and S8 If the rates of the S4 and S8 formation reactions behave as Babikov predicts, then each reaction should generate 500‰ fractionation (i.e., a factor of 2) 33S should be negative for elemental sulfur 36S should be positive and approximately twice as large Harman et al., Science (submitted)

MIF generation during formation of S4 and S8 In the linear regime, the predicted slope of 33S vs. 34S is 0.5, in agreement with the data In the linear regime, the slope of 36S vs. 33S is 2 Harman et al., Science (submitted)

New clues from 36S For large fractionations, one needs to use the more precise MIF formulae (from Farquhar et al., 2001) For 33S = 34S = 36S = 500‰, we get: 33S = -199‰ 36S = +230‰ so the slope of 36S/ 33S = +230‰/(199‰) = 1.16, in approximate agreement with what is observed

What happens when one puts this mechanism into a photochemical model? In a paper that is submitted to Science, with Harman, Pavlov, Babikov, and Kasting as coauthors, we did just that Base atmosphere assumptions: 1 bar N2 background atmosphere 3% CO2 (~100 PAL) H2 outgassing rate = 1×1010 cm-2s-1 Solar UV flux appropriate for 2.5 Ga

Results of the photochemical model calculation The calculations show that the MIF signal that is generated is a strong function of the combined SO2 and H2S outgassing rate The predicted fractionations are much higher than observed, but this could be explained by dilution of the MIF signal by biological/chemical reprocessing in the Archean marine biosphere Harman et al., submitted

Results of the photochemical model calculation At low sulfur outgassing rates, most of the sulfur is removed as H2S and sulfate, neither of which is strongly fractionated At higher outgassing rates, the dominant exit channel is S8, followed by sulfate, and everything besides S8 is strongly positive in 33S Harman et al., submitted

Comparisons with data The predicted model slope for 33S vs. 34S is ~0.5, which is only half that seen in the data This is for the nomimal model, with a sulfur outgassing rate equal to 1/3rd of today The slopes can be adjusted by tweaking various model parameters Data on 2.5-2.7 Ga sulfides (from Ono et al. (2003). Figure from Harman et al., submitted.

Comparisons with data The predicted model slope for 36S vs. 33S is approximately 1.5, whereas the data have a slope of about 0.9 The dark shaded region shows how the slope can be changed by adjusting reaction rate constants The light shaded region shows the result of adding a photosynthetic O2 flux at the surface Data on 2.5 Ga sulfides in W. Australia and S. Africa (from Kaufman et al., (2007) and Ono et al. (2009). Figure from Harman et al., submitted.

Comparisons with data Finally, our model may be able to explain the 33S ‘spike’ in the Late Archean Photosynthetic O2 would have created more sulfate, which has positive 33S If this sulfate was quickly converted to pyrite by bacterial sulfate reduction, then this signal could have been preserved S8 production must have continued in regions that were not influenced by O2 plumes D. T. Johnston, Earth Sci. Rev (2011)

Possible stumbling blocks for the hypothesis Barite shows negative 33S, whereas the theory predicts that atmospheric sulfate should have positive 33S This problem can be solved if biota in the ocean oxidized S8 to sulfate during anoxic photosynthesis or using photosynthetic O2

Possible stumbling blocks for the hypothesis Elemental sulfur produced in laboratory SO2 photolysis experiments has positive 33S, whereas our mechanism predicts that it should be negative However, the experiments are subject to self-shielding, which causes the major SO2 isotopologue to photolyze more slowly than its isotopic counterparts, producing positive 33S in S and S2

Possible stumbling blocks for the hypothesis If elemental sulfur condenses out on the walls of the relatively small reaction chamber as S or S2, which are not fractionated during chain formation, then it may retain its positive 33S signal

Conclusions Atmospheric O2 levels were low prior to ~2.45 Ga Cyanobacteria produced this first O2 The best evidence for this comes from sulfur MIF The large MIF signal is probably generated by multiple processes, of which sulfur chain formation is arguably the most important The spike in 33S between 2.5 and 2.7 Ga may be caused by the evolution of oxygenic photosynthesis Chemical reaction networks have the potential to generate MIF. This could have implications for other geochemical/isotopic systems