1. Serpentinization and the Great Oxidation Event Jim Kasting Dept. of Geosciences Penn State University

2. Dick Holland (1927-2012) Dick worked on many things during his career, but elucidating the details about the rise of O 2 was perhaps his greatest love, as well as his greatest scientific contribution When we got together, talk would invariably turn to the question: What caused the Great Oxidation Event (GOE) at 2.5 Ga? Dick’s passing spawned a special volume of Chemical Geology in 2013. Much of this talk is based on my own contribution to that volume

3. ‘Conventional’ geologic O 2 indicators Blue boxes indicate low O 2 Red boxes indicate high O 2 Dates have been revised; the initial rise of O 2 is now placed at ~2.45 Ga H. D. Holland (1994) Colorized by Y. Watanabe (Detrital)

4. Farquhar et al., Science (2000) 73 Phanerozoic samples High O 2 Low O 2 This story was strongly supported by sulfur isotope evidence published by James Farquhar and colleagues in 2000 So-called “mass independent fractionation” of S isotopes is seen before ~2.45 Ga, but not afterwards

5. Updated sulfur MIF record As S-MIF data have accumulated, the “cliff” at 2.45 Ga has become even more pronounced Small, but finite,  33 S values immediately after this may be caused by reworking of older sediments Alternative explanations cannot explain the corresponding MIF signal in 36 S Reinhard and Planavsky, Nature (2013) Grey circles—SIMS Open circles—bulk rock

6. Resolved and unresolved questions Thus, for many of us, the question of when atmospheric O 2 first rose (and stayed high) has been largely resolved –The answer is 2.45 Ga, and the event is often termed the Great Oxidation Event, or GOE But the question of why it rose at that time continues to be debated

7. What caused the GOE (Great Oxidation Event) at ~2.4 Ga? In one sense, the answer to this question is easy: The rise of O 2 was caused by cyanobacteria, the only true bacteria capable of performing oxygenic photosynthesis In another sense, though, the rise of O 2 is a mystery, as both cyanobacteria and oxygenic photosynthesis appear to predate the GOE by several hundred million years The best evidence for this comes from measurements of Mb in shales  http://www.primalscience.com/?p=424

8. Science (2007) Mb is forms an insoluble sulfide in reduced environments A Mb enhancement in shales requires oxidative weathering of sulfides on land, followed by transport of soluble molybdate ion to sediments The best way to do this, in my view (and that of Reinhard and Planavsky), is for the entire atmosphere to become O 2 -rich for short time periods

9. Additional evidence from Cr isotopes suggests that O 2 was being produced as far back as 3.0 Ga (Crowe et al., Nature, 2014) If one accepts these arguments for O 2 production during the Archean, the real question becomes: What delayed the rise of atmospheric O 2 ? – Obviously, something was holding O 2 concentrations down, but what exactly was doing so?

10. Published hypotheses for the cause of the GOE * 1.Progressive mantle oxidation (Kasting et al., 1993) 2.Holland’s tectonic evolution/volcanic outgassing model (Holland, 2002, 2009) 3.Submarine versus subaerial outgassing mechanisms (Kump and Barley, 2007; Gaillard et al., 2011) 4.Continental oxidation and hydrogen escape (Catling et al., 2001; Catling and Claire, 2005; Claire et al., 2006) 5.Serpentinization of seafloor (Kasting and Canfield, 2012) 6.Banded iron-formation triggers (Isley and Abbott, 1999; Barley et al., 2005; Goldblatt et al., 2006; Bekker et al., 2010) 7.Various biological triggers –Ni famine for methanogens (Konhauser et al., 2009) –Nitrogenase protection mechanisms; Mo/V availability (Anbar and Knoll, 2002; Grula, 2005; Zerkle et al., 2006; Scott et al., 2008, 2011; Kasting and Canfield, 2012) * See J. F. Kasting, Chem. Geol. (2013)

11. 1. Progressive mantle oxidation The idea here was that H escape to space oxidizes the upper mantle (because the H came from H 2 O originally) Volcanic gases therefore become more oxidized with time Some support for this hypothesis was provided by sulfide barometry in 3.3-3.5 Ga peridotitic diamonds, which suggested that the upper mantle was more reduced at that time Kasting et al., J. Geol. (1993)

12. Unfortunately, studies of Cr (J.W. Delano, 2001) and V (D. Canil, 1997, 2002; Li and Lee (2004) concentrations in ancient basalts and peridotites appear to have ruled out this hypothesis –These elements partition differently into the melt as a function of their redox state But the idea that one needs to get more H 2 out of the early Earth to delay the rise of O 2 remains valid

13. 2. Holland’s tectonic evolution/volcanic outgassing model During the last decade of his life, Dick Holland published two papers outlining his hypothesis for the cause of the rise in O 2 (Holland, GCA, 2002, 2009) In his view, the composition of volcanic gases was the key…

14. Holland’s f-value analysis (GCA, 2002) Holland analyzed the redox state of volcanic gases by calculating their f-values Volcanic gases with f >1 lead to a reduced environment A key assumption is that 20% of outgassed CO 2 is buried as organic C

15. The carbon isotope record  13 C carb = 0 corresponds to 20% organic carbon burial Except during times of transition, this is about what we see

16. Holland’s 2009 model In his 2009 GCA paper, Dick proposed an explicit model for determining the timing of the GOE Gradual growth of the continents resulted in increased recycling of C and S relative to H 2 O, and hence in lower f- values –So, volcanic gases do not become more oxidized, but their C/H and S/H ratios change If one picks parameters carefully, the GOE occurs at ~2.5 Ga Holland, GCA (2009 )

17. Big question with this analysis: What controls the organic carbon burial fraction in ancient sediments? – If the 20% organic C burial fraction is controlled by redox balance, as seems likely, then invoking this as a constraint on O 2 evolution involves circular reasoning – An alternative hypothesis is that the organic C burial fraction is controlled by the C:P ratio in igneous rocks (Junge et al., JGR, 1975), but this also seems unlikely

18. 3. Submarine/subaerial outgassing and the GOE Two different papers have argued that the rise of atmospheric O 2 at ~2.4 Ga was caused by a switch from predominantly submarine to predominantly subaerial volcanism Kump and Barley outlined the geologic evidence for this switch and analyzed data from submarine and subaerial volcanic gases Kump and Barley, Nature (2007)

19. Hypothesis: The rise in atmospheric O 2 was caused by a switch from submarine to subaerial volcanism f was >1 in the Archean and <1 afterwards, in their view Much of the change in f, however, is driven by the implicit assumption of 20% organic C burial –Hydrothermal fluids are rich in H 2 S compared to CO 2 More reduced Nature (2007)

20. Submarine/subaerial outgassing and the GOE Gaillard et al. (Nature, 2011) performed detailed modeling of the outgassing process and emphasized the switch from submarine outgassing of H 2 S to subaerial outgassing of SO 2 Gaillard et al., Nature (2011)

21. In none of Gaillard’s cases did Holland’s f value ever exceed 1 for submarine outgassing (primarily because CO 2 outgassing increases at high pressures, leading to more organic carbon burial)  This mechanism doesn’t seem to work It does somewhat better if one uses an alternative approach to analyzing the global redox budget (see below), but still is not a sufficient answer Nature (2011) Subaerial outgassing Submarine outgassing

22. Before discussing the remaining hypotheses for triggering the GOE, we need to step back and talk about how best to quantify the principle of redox balance –Most of this discussion follows J.F. Kasting, Chem. Geol. (2013)

23. The principle of redox balance It’s not just a good idea, it should be a law… Two ways to think about it 1.Conservation of free electrons 2.When one thing is oxidized, something else must be reduced Although this methodology for analyzing the GOE differs from Holland’s f-value analysis, all of Holland’s papers emphasized global redox balance, so this way of thinking is definitely not new

24. Two important redox budgets The atmospheric redox budget (at left) must be balanced for low-O 2 atmospheres. This is how we formulate our Archean photochemical models The global redox budget (at right) must be balanced for all atmospheres. This is the budget of the combined atmosphere-ocean system Combined atmosphere- ocean system

25. Redox budget formulation Define “neutral” oxidation state gases: H 2 O, CO 2, N 2, and SO 2 Other gases are either oxidized or reduced compared to these. Express the differences in terms of H 2 equivalents, e.g. CH 4 + 2 H 2 O  CO 2 + 4 H 2 H 2 O 2 + H 2  2 H 2 O Thus, the total outgassing rate of reductants can be written as

26. Global redox balance equation Setting H 2 sources equal to H 2 sinks yields the following equation: Here  out (Red) = total outgassed flux of reduced gases  OW = oxidative weathering of the continents and seafloor  burial (CaSO 4 ) = burial of gypsum or anhydrite  burial (Fe 3 O 4 ) = oxidation of ferrous iron without using O 2 (includes BIFs and serpentinization)  burial (CH 2 O) = burial of organic matter  burial (FeS 2 ) = burial of pyrite

27. Catling & Claire’s * K oxy parameter Let K oxy represent the ratio of H 2 sinks to sources, not counting oxidative weathering and burial of sulfate (which are only important at high O 2 ) and escape of hydrogen to space (which is only important at low O 2 ) The atmosphere switches from reduced to oxidized when K oxy becomes >1 * Catling and Claire, EPSL (2005) Claire et al., Geobiology (2006)

28. Differences between my analysis and Catling & Claire’s approach 1.Budgeting done in terms of H 2 rather than O 2 2.I take SO 2 as the neutral oxidation state for sulfur, whereas most authors, including Holland, used sulfate –S was neglected by Claire et al. (2006) 3.I lump metamorphic and volcanic fluxes together 4.I include anaerobic iron oxidation (e.g., BIF deposition and serpentinization) in my model

29. Magnitudes of terms Relative magnitudes today (mostly from Dick Holland’s work): TermRate (10 12 mol/yr) 2  burial (CH 2 O) 20 ± 6.6 5  burial (FeS 2 ) 7 * ± 4  out (red)4.8 ± 3.6  burial (Fe 3 O 4 )0.4 ± 0.2  K oxy  5.2 * Corrected because of mistake in Holland (2002), Tables A1 and A2

30. This is a big problem, because we need K oxy < 1 during the Archean to maintain a reduced atmosphere –It helps if one uses Berner’s organic carbon burial rate, which is a factor of 2 smaller, but then one needs to rebalance all of the fluxes to remain consistent with the carbon isotope record

31. 4. Catling & Claire’s continental oxidation model In the models of Catling et al. (2001), Catling and Claire (2005), and Claire et al. (2006), loss of H to space oxidized the continents, resulting in a smaller flux of reduced metamorphic gases There is indeed evidence that much of the oxygen stored in continental rocks was emplaced before the GOE (see diagram at right) Catling et al., Science (2001)

32. Catling and Claire model Modern metamorphic reduced gas fluxes were scaled up by factors of 20-50 using thermodynamic equilibrium arguments But –The continents may have been much smaller back then –Thermodynamic equilibrium does not apply at metamorphic temperatures –Most of the modern metamorphic reduced gas flux is CH 4. And most of that CH 4 is thermogenic, i.e., it is produced from heating of buried organic matter. Only about 20% of it is abiotic, produced by serpentinization  Catling and Claire may have overestimated the metamorphic flux of hydrogen during the Archean

33. Serpentinization One needs a kinetic mechanism for oxidizing rocks without using O 2 or sulfate Serpentinization is one such mechanism that operates today Ultramafic rocks interact with warm water to form serpentine minerals, producing hydrogen in the process Serpentine cabochon from China. This is approximately 39 millimeters by 23 millimeters (From Geology.com)

34. Serpentinization Iron is excluded from the serpentine minerals, so it goes into magnetite 3 FeO + H 2 O  Fe 3 O 4 + H 2 This mechanism is aided by the fact that Archean continental rocks were observably more ultramafic (greenstone belts, komatiites) Serpentine cabochon from China. This is approximately 39 millimeters by 23 millimeters (From Geology.com)

35. 5. Serpentinization of seafloor But, if the continents contained a lot of ultramafic rock that was prone to serpentinization, wouldn’t the seafloor have been undergoing serpentinization, as well? Today, most water-rock interactions occur within the midocean ridge hydrothermal vents This was probably even more true during the Archean, especially if the continents were smaller

36. EPSL, 2010 The Archean mantle would have been hotter, leading to a higher degree of partial melting at the midocean spreading ridges Definition: Convective Urey ratio = heat generated by radioactive decay within the mantle/mantle convective heat flow Blue curves represent thermal evolution models for different present-day convective Urey ratios (from Korenaga, 2008)

37. EPSL, 2010 The Archean mantle would have been hotter, leading to a higher degree of partial melting at the midocean spreading ridges More melting makes the resulting igneous rock more like the mantle, which is rich in Fe and Mg Such models also predict very thick oceanic crust, which would cool slowly, possibly giving rise to widespread hydrothermal circulation Modern seafloor: 10-13 wt% MgO Archean seafloor: 18-24 wt% MgO

38. A recent paper based on a statistical analysis of ~70,000 major and trace element measurements of various continental rocks supports the idea that the early crust was ultramafic According to these authors, the percentage of fractional melting during volcanism has decreased from ~35% in the Archean to ~10% today A sharp decrease in fractional melting occurred right near the Archean-Proterozoic boundary This supports the idea that more serpentinization, and hence more H 2 production, was occurring during the Archean Keller and Schoene, Nature (2012)

39. Conclusions None of the published mechanisms provides a fully satisfactory explanation for the GOE –However, elements of several mechanisms are probably part of the story. In particular, Earth’s tectonic evolution is important to consider The f-value analysis is circular because the assumption of 20% organic C burial is hard-wired into it. The flux balance approach of Catling and Claire (2006) and Kasting (2013) avoids this circularity Changes in the composition and thickness of the oceanic crust may well have been the most important factor in triggering the GOE. How can we better quantify this?