Production of hydrogen peroxide in the atmosphere of a Snowball Earth and the origin of oxygenic photosynthesis

Abstract

During Proterozoic time, Earth experienced two intervals with one or more episodes of low-latitude glaciation, which are probable “Snowball Earth” events. Although the severity of the historical glaciations is debated, theoretical “hard Snowball” conditions are associated with the nearly complete shutdown of the hydrological cycle. We show here that, during such long and severe glacial intervals, a weak hydrological cycle coupled with photochemical reactions involving water vapor would give rise to the sustained production of hydrogen peroxide. The photochemical production of hydrogen peroxide has been proposed previously as the primary mechanism for oxidizing the surface of Mars. During a Snowball, hydrogen peroxide could be stored in the ice; it would then be released directly into the ocean and the atmosphere upon melting and could mediate global oxidation events in the aftermath of the Snowball, such as that recorded in the Fe and Mn oxides of the Kalahari Manganese Field, deposited after the Paleoproterozoic low-latitude Makganyene glaciation. Low levels of peroxides and molecular oxygen generated during Archean and earliest Proterozoic non-Snowball glacial intervals could have driven the evolution of oxygen-mediating and -using enzymes and thereby paved the way for the eventual appearance of oxygenic photosynthesis.

Hydrogen peroxide provides a powerful oxidant in anoxic environments. The recent discovery of H₂O₂ in the atmosphere of Mars (1, 2) supports the hypothesis that oxidation of iron by photochemically generated H₂O₂ over the past 4 billion years may have yielded the present oxidized Martian surface (3). The cold, dry, and low-oxygen (≈7 μbar) Martian atmosphere concentrates H₂O₂ production near the surface (4, 5). Hydrogen peroxide can be preserved readily in ice, because H₂O₂ has a slightly lower freezing point (−1°C) than H₂O; a concentration of H₂O₂ as high as 0.13% is observed on the surface of the Galilean satellite Europa (6). The closest analog in Earth history for the cold, dry, low-oxygen conditions of Mars and Europa may have occurred during the proposed Paleoproterozoic Snowball Earth event (7, 8) and perhaps during one or both of the proposed Neoproterozoic Snowball Earth events (9, 10).

Although there have been many glacial events recorded in the history of the Earth (9–11), two major periods of low-latitude glaciation in the Proterozoic appear correlated with significant changes in the evolution of life (8, 12–14) and atmospheric oxygen level (15–19). The Paleoproterozoic Makganyene glaciation occurred approximately between 2.3 and 2.2 Ga, and at least two other low-latitude glaciations occurred during the Cryogenian period, between ≈740 and 630 Ma (9, 10, 20). The severity of these “Snowball Earth” events is debated, but the low latitude of the glaciations indicates that, at least on the continents, ice extended to the equator, average global temperatures were likely well below freezing, and the hydrological cycle was much diminished (21, 22). The rock record indicates that the atmosphere and ocean were oxygen-poor until shortly before the onset of the Paleoproterozoic Snowball at ≈2.3 Ga (15–19), and the weakening of the biosphere and hydrological cycle would likely have decreased atmospheric oxygen levels during the event. Here we adopt a photochemical model similar to the one that Nair et al. (4) successfully applied to the Martian atmosphere to investigate the production and deposition of H₂O₂ during low-oxygen Snowball Earth conditions.

We examined atmospheric photochemistry during the Paleoproterozoic Snowball Earth event, as well as the impact of earlier glaciations on the evolution of cellular life. Assuming the Paleoproterozoic Makganyene glaciation was a “hard Snowball,” it led to an environment similar to that on Mars and icy satellites, where H₂O₂ could be produced and preserved in ice. During the deglaciation, the deposited H₂O₂ would have been released into the oceans and atmosphere, as occurs during the spring and summer in Greenland and at the South Pole (23, 24). On a low-oxygen planet, the H₂O₂ could have provided an important source of oxidants for driving the evolution of oxygen-mediating and -using enzymes.

Atmospheric Chemical Models

We performed a one-dimensional diurnally averaged simulation of the chemical processes in the atmosphere of a Snowball Earth. Our model calculates the profiles of O, O(¹D), O₂, O₃, H, H₂, OH, HO₂, H₂O₂, CO, CO₂, HCO, and H₂CO by solving the mass continuity equation (∂n_i/∂t) + (∂ϕ_i/∂z) = P_i − L_i, where n_i is the number density for species i, ϕ_i is the vertical flux, P_i is the chemical production rate, and L_i is the chemical loss rate, all evaluated at time t and altitude z. P_i and L_i are calculated based on the chemical schemes published in the literature (4, 25–27). The vertical flux is given by

where D_i is the species' molecular diffusion, H_i is the species' scale height, H_atm is the atmospheric scale height, α_i is the thermal diffusion parameter, and T is the temperature. The vertical eddy mixing coefficient K_zz in this work is set to be 2 × 10⁵ cm²·s⁻¹, which is close to the present-day value near the surface (25). With this K_zz, the vertical mixing time is on the order of 10⁶ s, a value much longer than the lifetime of H₂O₂ of ≈10⁴ s in the atmosphere. We therefore expect that K_zz plays a minor role in the vertical profile of H₂O₂.

Because of model limitations, we assume the surface pressure is 1 bar over the course of model time. We fix the H₂O profile at that determined by its saturation pressure in the atmosphere; we vary the profile by modifying the surface temperature. For the temperature profile, the surface temperature is taken to be 240 K, with a vertical temperature gradient of −10 K·km⁻¹ (dry-adiabatic lapse rate) and a constant temperature of 150 K in regions where the extrapolation of the temperature with the assumed gradient yields values <150 K; this is quoted as the reference model. We also assume a present-day solar UV spectrum. Initially, the atmosphere contains N₂ and H₂O only. The model starts with an upward H₂ and CO₂ flux of 10¹⁰ molecules cm⁻²·s⁻¹, which are close to the current volcanic outgassing rates (13). Species other than H₂, H₂O, and H₂O₂ are impermeable at the boundary. We allow H₂ to escape to space hydrodynamically when the conditions of Tian et al. (28) are met, although the strength of hydrodynamic escape in this paleoatmosphere remains inconclusive (29). The model results with both higher and lower levels of hydrogen escape are also presented. With the present-day solar spectrum, hydrodynamic escape happens when the homopause (≈65 km) mixing ratio of H₂ is >0.05. To a first-order approximation, we assume that H₂ escape with a flux of 10¹⁰ molecules cm⁻²·s⁻¹ is initiated when the H₂ mixing ratio is >0.1 and that it remains at this level as long as the ratio is >0.05. We let the model run for the entire lifetime of the Snowball (see below). Vertical profiles of species important to this study are summarized in Fig. 1; some other profiles are presented in Fig. 2. Note that formaldehyde, which can be preserved on the surface by precipitation (27), is not produced in large amounts, so the deposition of H₂ by burying H₂CO is insignificant.

Fig. 1.

Profiles of H₂O (dotted line), H₂O₂ (solid line), O₂ (dash-dotted line), H₂ (triple dot-dashed line), CO (long-dashed line), and CO₂ (dashed line) calculated with the reference model, in which surface temperature is 240 K, temperature gradient is −10 K km⁻¹, and constant temperature is 150 K above the tropopause (≈10 km). The values are shown for the model calculation at one instance at the end of the Snowball Earth. The downward and escape fluxes of H₂O₂ and H₂ are ≈5 × 10⁸ and 10¹⁰ molecules cm⁻²·s⁻¹, respectively. The mixing ratio of H₂O above the tropopause is set equal to that at the tropopause.

Fig. 2.

Profiles of H (solid line), O (dotted line), O₃ (dashed line), HCO (dash-dotted line), and H₂CO (triple dot-dashed line) under the same conditions as Fig. 1.

Current thinking suggests that deglaciation would proceed from build-up of greenhouse gases such as CO₂ due to persistent volcanic activity uncompensated by silicate weathering. As ice cover and low temperatures would have greatly hindered silicate weathering, atmospheric accumulation of greenhouse gases, particularly CO₂, would have eventually allowed temperatures to rise above freezing (30). In Neoproterozoic time, over the likely 12-million-year lifetime of the Snowball event (20), atmospheric CO₂ could have reached a level of 0.2 bar at the present-day outgassing rate of ≈10¹⁰ molecules cm⁻²·s⁻¹. Although it has recently been noted that greenhouse warming caused directly by CO₂ at a level as high as 0.2 bar may not have been sufficient to initiate the deglaciation of Neoproterozoic Snowballs (30, 31), in our model we still follow typical suggestions to estimate the duration of the Paleoproterozoic Snowball event. Assuming 0.2 bar CO₂ was actually required to start the Neoproterozoic deglaciations, the lower solar constant in the Paleoproterozoic (85% of modern insolation, rather than 96%) implies that triggering the end of a Paleoproterozoic Snowball would have required ≈0.6 bar CO₂. At present degassing rates, such an amount would have taken ≈35 million years to build up (32).

Hydrogen Peroxide Deposition

Hydrogen peroxide can be deposited on ice by diffusional contact. This deposition rate is limited by the lifetime of H₂O₂ against photolysis in the atmosphere (≈6 h), equivalent to a traveling distance of <1 km for a diffusion coefficient of 2 × 10⁵ cm⁻²·s⁻¹. The diffusional deposition of H₂O₂ is also sensitive to the H₂/O₂ ratio (Fig. 3). The abundance of H₂ reflects the strength of hydrogen escape to space. With a ratio of ≈2 the deposition rate is as high as 10¹⁰ molecules cm⁻²·s⁻¹. For ratios >5, the rates are suppressed by a factor of >10. The accumulation of H₂ in the atmosphere would soon stop the diffusional deposition of H₂O₂ if the replenishment of O₂ were insufficient to support an O₂ abundance as high as that in Fig. 1, because of the reaction 2H₂ + O₂ → 2H₂O. We find that the continuous accumulation and photolysis of CO₂ in the atmosphere can maintain such a level of O₂ (2CO₂ → 2CO + O₂) via well known HOx chemistry (26). Consequently, CO is in high abundance (Fig. 1). Because the photolytic processes of CO₂ and H₂O are in similar wavelength ranges, by the end of the Snowball CO₂ becomes a major UV absorber and H₂O photolysis is less significant. In this standard model, the deposition rate of H₂O₂ by diffusion is 3 × 10⁸ molecules cm⁻²·s⁻¹.

Fig. 3.

Sensitivity of surface H₂ abundance, obtained by varying the strength of hydrogen hydrodynamic escape, on the deposition rate of H₂O₂ by diffusion. The abundance of O₂ as high as that in Fig. 1 is maintained by CO₂ photolysis (see text). The rightmost point is calculated by assuming that H₂ escape vanishes. For the diffusion-limit hydrogen escape case, the surface H₂ mixing ratio is ≈5 × 10⁻⁴ and the H₂O₂ deposition rate is ≈4 × 10⁸ molecules cm⁻²·s⁻¹.

For the H₂O₂ precipitation rate, we follow the method developed for modern glacial environments (23, 24) to estimate a concentration of H₂O₂ in snowfall of ≈20 μM. Incorporating the estimated H₂O₂ diffusional flux to surface ice and assuming a hydrological cycle of f mm·yr⁻¹, this standard model yields a volume mixing ratio of H₂O₂ in the snow/ice as high as ≈3 × 10⁻⁶(1/f) if f < 10, and as low as ≈5 × 10⁻⁷ if f > 10. Enhancing the hydrological cycle will always enhance the rainout rate of H₂O₂ as well as the concentration of H₂O₂ in H₂O ice, by analogy with evaporation processes in H₂O₂-H₂O solution, which enhance the concentration of H₂O₂ in the solution (33). This further concentration of H₂O₂ in the ice depends on the partitioning of H₂O₂ and H₂O in vapor during evaporation (knowledge of which is not available for low-temperature conditions), the lifetime of ice sheet (34), and the strength of hydrological cycles or precipitation/evaporation rates (31, 34). A one-dimensional model of the dynamics of ice on a Snowball Earth has been published (34).

Changing surface temperature will also modify the calculated deposition rate of H₂O₂ by diffusional contact. Reducing surface temperature will correspondingly reduce the atmospheric H₂O vapor abundance and hence move the oxidation line (the region of the atmosphere where H₂O₂ is mostly produced) closer to the surface. This shift will enhance diffusional H₂O₂ deposition, because the lifetime of H₂O₂ is short compared with vertical transport time (10⁴ vs. 10⁶ s; see above). The results of tests of sensitivity to changes of surface temperature are summarized in Fig. 4. In general, reducing surface temperatures will enhance the deposition rate by diffusion but will reduce the rainout rate.

Fig. 4.

Sensitivity of surface temperature on the deposition rate of H₂O₂ by diffusion. The vertical profiles of major compounds (e.g., H₂, O₂, CO, and CO₂) are similar to those in Figs. 1 and 2.

Direct rainout from H₂O₂ nucleation in the region between ≈10 and 20 km is not included in the current model; the partial vapor pressure of H₂O₂ is ≈5 × 10⁻⁸ mbar, which is greater than the saturation pressure (33) of ≈10⁻⁹ mbar at 150 K. If this amount of H₂O₂ is removed by rainout, an additional precipitation rate of ≈10¹⁰ molecules cm⁻²·s⁻¹ (equal to the production rate in this region) is imposed, yielding the total volume mixing ratio of H₂O₂ in the ice as high as 10⁻⁴(1/f). Note that the total production rate of H₂O₂ by gas-phase chemistry is ≈10¹¹ molecules cm⁻²·s⁻¹, which provides an upper limit to the rate of H₂O₂ deposition by diffusion and precipitation.

We stress that the calculated concentration of H₂O₂ in the snow/ice is a lower limit. With the present-day solar spectrum, the maximum photolysis rate of H₂O is ≈10¹² molecules cm⁻²·s⁻¹. If UV photons directly irradiated the snow/ice, H₂O₂ could be produced and preserved readily in the ice. For example, the production rate of H₂O₂ in Antarctic seawater samples by direct sun light irradiation has been shown to approach this maximum H₂O photolysis rate (35). Recently, it has also been shown that H₂O₂ at the surface is enhanced when stratospheric ozone is depleted (36), as in an anaerobic atmosphere. This mechanism of depositing H₂O₂ would be insensitive to the abundance of atmospheric H₂.

In summary, we find that the volume mixing ratio of H₂O₂ in the ice falls between 5 × 10⁻⁷ and 10⁻⁴(1/f). The lower limit represents the regions where water cycles (precipitation/evaporation and ice melting) are active, as might occur at midday, in the summer, and in low-latitude regions. The upper limit can be achieved where water cycles are weak, such as in mid to high latitudes of the winter hemisphere (31). For example, at low latitudes of the summer hemisphere, the water precipitation could be as high as 45 cm·yr⁻¹ (31), resulting in the mixing ratio of H₂O₂ in the ice ≈5 × 10⁻⁷. At winter hemisphere mid to high latitudes, pure H₂O₂ rainout is possible, because of low water cycles (31). Coupled with sea ice flow (34), a high level of H₂O₂ ice formed at mid to high latitudes can be transported to a lower latitude. Consequently, high-concentration H₂O₂ ice is available globally on the Snowball Earth. (A volume mixing ratio of H₂O₂ of 10⁻³ stored in 1-km-thick ice could disproportionate to form the equivalent to 0.1 bar of O₂.) We note that the above estimation is simply based on snowfalls and dry deposition. The direct freezing of sea water could affect the concentration of H₂O₂ in the frozen ice; however, this process is sensitive to the strength of dynamical mixing in sea water. High concentrations of H₂O₂ in the ice may significantly suppress hydrological cycles. Future work is needed to resolve the aforementioned uncertainties.

Biological Implications

Even when the concentration of H₂O₂ in snow and ice is as low as that in modern polar regions, the release of H₂O₂ into the ocean upon melting could provide an environmental oxidant that would threaten organisms living nearby. Hydrogen peroxide in the presence of ferrous ion produces the hydroxyl radical (and ferryl iron), which is lethal to the cell (37). The Mn-based enzyme catalase, which catalyzes the reaction 2H₂O₂ → 2H₂O + O₂ (38, 39), and the superoxide dismutase enzymes, which neutralize O₂⁻, protect the cell against the effects of hydrogen peroxide and the hydroxyl radical. These enzymes likely evolved before the evolution of oxygenic photosynthesis and hence protected the first oxygen-producing phototroph (40), perhaps in response to an environmental peroxide challenge. Blankenship and Hartman (41) further suggested that H₂O₂ played a crucial role in the origin and evolution of oxygenic photosynthesis because it is capable of being both a powerful oxidant and a reductant and because the oxidation of H₂O₂ to O₂ is fully within the oxidative capabilities of reaction centers from existing anoxygenic photosynthetic bacteria (41). The Huronian glaciations at ≈2.4–2.3 Ga (8), the Pongola glaciations at ≈2.9 Ga (42), and perhaps unrecognized earlier glacial episodes might thus have spurred their development of both oxygen tolerance and oxygenic photosynthesis (40, 41), as well as stimulated the evolution of diverse oxidase and peroxidase enzymes that are now critical for aerobic metabolism.

Kopp et al. (8) proposed that the severity of the Makganyene Snowball Earth event was caused by the evolution of efficient oxygenic photosynthesis during the short interval between the Huronian glaciations and the Makganyene glaciation, and that photosynthetic oxygen production at the onset of the Makganyene glaciation triggered a collapse of the methane greenhouse. The evolution of oxygen tolerance in response to peroxide build-up during the Huronian glaciations could have paved the way for this evolutionary step.

Only a planetary glaciation like a Snowball Earth event, however, would be likely to produce enough H₂O₂ to leave a global fingerprint. During a Snowball, as ocean waters cycled through hydrothermal vents, the concentration of metals like Fe²⁺ and Mn²⁺ would have built up to high levels. After the Snowball, as recorded in the Kalahari Manganese Field (7) and in lesser Neoproterozoic manganese deposits like those of the Urucum district (43), iron and manganese would have been oxidized by oxygen and precipitated out of solution. After the Neoproterozoic Snowballs the main source of oxygen was likely atmospheric, but after the Makganyene Snowball atmospheric oxygen was still fairly low. The oxygen source for the Kalahari Manganese Field was probably a combination of a post-Snowball cyanobacterial bloom (7) and disproportionation of H₂O₂.

Fennel et al. (44) recognized recently that nitrate limitation is a critical problem in the transition from an anaerobic to an aerobic environment, such as presumably occurred around 2.3 Ga at the “Great Oxygenation Event.” Starting from an anaerobic environment, increasing oxidation removes metal cofactors critical for the function of the nitrogenase enzyme, as well as enabling aerobic denitrification. In turn, nitrogen limitation acts to constrict productivity, keeping the global production of oxygen by the cyanobacteria at levels below those needed to transition into the modern, oxygen-dominated stable environment in which abundant nitrate is biologically available. Fennel et al. (44) offer no clear explanation of how this “biogeochemical bottleneck” barrier between the anaerobic and aerobic stable states of the planetary ecosystem could be crossed. We suggest tentatively that the post-Makganyene snowball “burp” of peroxide (which in some model scenarios could be on the order of 1 bar) might have pushed the global environment over this nitrate limitation barrier by throwing the surface ocean into the oxic, nitrate-rich realm. More sophisticated modeling is necessary to test this idea.

We have demonstrated here that the generation of H₂O₂ through photolytic processes involving H₂O is possible in the oxygen-poor early atmosphere of the Archean and early Paleoproterozoic (45, 46). Large amounts of hydrogen peroxide can be preserved only under special conditions, such as during an intense glaciation. Because of the short atmospheric lifetime of H₂O₂ against UV photolysis, build-up of large amounts requires H₂O₂ production close to the ground (hence, under low-O₂ conditions), as well as a mechanism to trap it in the ice to store and concentrate it. During an intense glaciation, particularly a Snowball event, the H₂O₂ deposition rate may have been enhanced greatly, as we discuss above, although our model is extremely simple and excludes many important considerations. A coupled chemistry (gas-phase and solid-state) and dynamics (climate and ice-sheet) model is needed to quantify our calculation in detail and to provide a more accurate and quantitative estimation of the deposition of hydrogen peroxide during glaciations in anaerobic atmospheres as well as Snowball Earth events.