ABIOTIC OXYGEN-DOMINATED ATMOSPHERES ON TERRESTRIAL HABITABLE ZONE PLANETS

The rapid growth of exoplanet discovery and characterization over the last two decades has fueled hopes that in the relatively near future, we may be able to observe the atmospheres of Earth-like planets spectroscopically. Such targets will be intrinsically interesting for comparative planetology, but also for the major reason that they may host life. To search for life on exoplanets by observing their atmospheres, we must first decide on spectral features that can be used as biosignatures. Despite extensive theoretical study of various possibilities, detections of molecular oxygen (O₂) and its photochemical byproduct, ozone (O₃), are still generally regarded as important potential indicators of Earth-like life on another planet (Segura et al. 2005; Kaltenegger et al. 2010; Kasting et al. 2013; Snellen et al. 2013).

Various authors have investigated the idea that abiotic oxygen production could lead to "false positives" for life (Selsis et al. 2002; Segura et al. 2007; Léger et al. 2011; Hu et al. 2012; Tian et al. 2014). For example, it has recently been argued that the build-up of O₂ to levels of ∼2–3 × 10⁻³ molar concentration in CO₂-rich atmospheres could occur for planets around M-class stars, because of the elevated XUV/NUV ratios in these cases (Tian et al. 2014). Extensive atmospheric O₂ buildup due to H₂O photolysis followed by H escape may also occur on planets that enter a runaway greenhouse state (Ingersoll 1969; Kasting 1988; Leconte et al. 2013). However, because by definition the runaway greenhouse only occurs on planets inside the inner edge of the habitable zone, it should not lead to identification of false positives for life.

For planets inside the habitable zone, it is commonly believed that H₂O photolysis will always be strongly limited by cold-trapping of water vapor in the lower atmosphere. The purpose of this note is to point out that a mechanism for O₂ build-up to levels where it is the dominant atmospheric gas exists for terrestrial¹ planets in the habitable zone around any star type. The reason for this is that the extent of H₂O cold-trapping depends strongly on the amount of non-condensable gas in the atmosphere.

Previously, we have shown that the degree to which a condensing gas such as H₂O is transported to a planet's upper atmosphere is determined primarily by the dimensionless number $\mathcal {M} = \epsilon p_v L / p_n c_p T_s$, where L is the specific latent heat of the condensing gas, c_p is the specific heat capacity at constant pressure of the non-condensing gas (or gas mixture), T_s is temperature, p_v and p_n are respectively the partial pressures of the condensing and non-condensing gases in the atmosphere, $\epsilon = m_v/ m_n$ is the molar mass ratio between the two gases, and all values are defined at the surface. $\mathcal {M}$ is essentially the ratio of the latent heat of the condensing gas (here, H₂O) to the sensible heat of the non-condensing gas (primarily N₂ on Earth; Wordsworth & Pierrehumbert 2013). Values of $\mathcal {M}>1$ ($\mathcal {M}<1$) correspond in general to situations where the upper atmosphere is moist (dry).

Figure 1 shows the surface temperature dividing the moist and dry upper atmosphere regimes as a function of p_n for a pure N₂ − H₂O mixture. As can be seen, on a planet with 1 bar of N₂, a surface temperature of >340 K is required for a moist upper atmosphere, in rough agreement with detailed radiative–convective calculations (Wordsworth & Pierrehumbert 2013). However, the required surface temperature is a strong function of p_n. For 0.1 bar only ∼295 K is required, while for 0.01 bar the value drops to ∼255 K. In general there is no reason to expect that Earth's atmospheric nitrogen inventory is typical for a rocky planet: in the inner solar system alone, the range of atmospheric N₂ as a function of planetary mass spans 3.3 times (Venus) to 6.6 × 10⁻⁴ times (Mars) that of Earth. Delivery and removal of volatiles on terrestrial planets is dependent on an array of complex, chaotic processes, so wide variations in inventories should be expected (Raymond et al. 2006; Lichtenegger et al. 2010; Lammer et al. 2009).

Zoom In Zoom Out Reset image size

The O₂ buildup mechanism can easily be understood intuitively by a thought experiment involving a hypothetical planet with a pure H₂O composition (Figure 2). Lacking atmospheric N₂, Ar, and CO₂, such a planet will initially have a pure H₂O atmosphere, with the surface pressure determined by the Clausius–Clayperon relation (Andrews 2010; Pierrehumbert 2011). If the planet has the same orbit and incident stellar flux as present-day Earth, it will most likely be in a snowball state (Budyko 1969). However, because H₂O cannot be cold-trapped when it is the only gas in the atmosphere, it will be photolyzed by XUV and UV radiation from the host star (primarily via H₂O + hν → OH* + H*). The resultant atomic hydrogen will escape to space at a rate dependent on factors such as the XUV energy input and the temperature of the thermosphere, and hence the atmosphere will oxidize.²

Zoom In Zoom Out Reset image size

In the one-dimensional limit with no surface mass fluxes, atmospheric O₂ will build up on such a planet until p_n is high enough to cold-trap H₂O and reduce loss rates to negligible values. In three dimensions, the initial atmospheric evolution may depend on the planet's orbit and sub-surface heat flux/transport rate, because on a tidally locked, ice-covered planet with pure H₂O atmosphere, conditions on the dark side could be so cold that even O₂ would condense. However, on a planet with Earth-like rotation and obliquity, all regions of the planet receive starlight at some point in the year, so once the surface O₂ inventory passed a given threshold, buildup of an O₂ atmosphere would likely be inevitable. In addition, for any planet, transient heating events such as meteorite impacts would be able to force transitions to a stable state of high atmospheric pressure.³

What about more general scenarios? First, we can relax the assumption of zero downward flux at the surface and consider cases where the created O₂ can be used to oxidize the interior. Then, redox balance dictates that atmospheric oxygen levels must build up until the loss of hydrogen to space is balanced by the surface removal rate of oxidizing material. For example, if an O₂ removal rate⁴ of 5 × 10⁹ molecules cm⁻² s⁻¹ at the surface is balanced by diffusion-limited H₂O loss, given an escape rate $\Phi = b_{\rm H_2O-O_2}f_{\rm H_2O}$ $(H_{\rm O_2}^{-1}-H_{\rm H_2O}^{-1})$, the molar concentration of H₂O at the cold trap⁵ must be 3 × 10⁻³ mol mol⁻¹ under Earth gravity. Here $b_{\rm H_2O-O_2}$ is the binary diffusion coefficient of H₂O in O₂ (Marrero & Mason 1972), $H_{\rm O_2}$ and $H_{\rm H_2O}$ are respectively the atmospheric scale heights of O₂ and H₂O, and $f_{\rm H_2O}$ is the cold trap H₂O molar concentration.

The surface O₂ partial pressure required to match this cold-trap concentration, which can be calculated by integrating the moist adiabat equation (Ingersoll 1969) as in Wordsworth & Pierrehumbert (2013), depends on both the surface and cold-trap temperatures. In Earth's present-day oxygen-rich atmosphere, the cold trap occurs at a relatively high T_t ∼ 210 K, due primarily to the warming effect of ultraviolet solar absorption by O₃ (Andrews 2010). Given T_s = 288 K and T_t = 210 K, $f_{\rm H_2O}$ = 3 × 10⁻³ mol mol⁻¹ requires a surface O₂ partial pressure of 0.15 bar. For a snowball planet with T_s = 240 K, this would drop to 0.022 bar. By comparison, for T_s = 288 K and T_t = 140 K, 0.025 bars is required.⁶ Because O₂ build-up should lead to O₃ formation and hence stratospheric heating, O₂ partial pressures of at least a fraction of a bar appear plausible once the planet's atmosphere reaches a steady state.

How would things change on a more complex planet where other atmospheric constituents were present? First, if the atmosphere contains some N₂ or Ar, the amount of O₂ required to block H₂O escape will clearly be decreased, and increased horizontal heat transport would reduce the likelihood of atmospheric bistability via O₂ condensation in the planet's regions of low surface instellation. Reduced gases such as methane, which can be outgassed from a planet's interior by abiotic processes (Levi et al. 2013; Guzmán-Marmolejo et al. 2013), could have lifetimes similar to those on Earth today in an O₂-rich atmosphere, although variations in O₃ and NO_x concentrations as a function of UV levels and atmospheric composition might alter this (Wayne 2000). In addition, volcanically emitted sulphur species and heterogenous chemistry will also affect the atmospheric redox balance. Future investigations using photochemistry models will allow constraints on the importance of these effects as a function of the water loss rate.

Surface/interior redox exchanges are another source of complexity on a low-N₂ Earth-like planet. If the planet forms with a hydrogen envelope that is lost to space early on (e.g., Genda & Ikoma 2008), its crust and oceans should initially be reducing, and the oxidized products of H₂O photolysis might react rapidly with the surface at first. However, as long as this occurred, the upper atmosphere would remain H₂O-rich and rapid photolysis could continue. Over time, the planetary surface and interior would become oxidized, decreasing their ability to act as an oxygen sink. Assuming Earth's present-day XUV flux, a lower limit on H₂ escape from a hydrogen-rich homopause is ∼4 × 10¹⁰ molecules cm⁻² s⁻¹ (Tian et al. 2005). Given this, an N₂-poor Earth could lose 2.1 × 10²² moles of H₂O over4 Gy, or 28% of the current ocean volume.⁷ This translates to 66.2 bar of atmospheric O₂—a large enough quantity to cause significant irreversible oxidation of the solid planet and hence a strong decrease in the reducing power of the surface. Because XUV fluxes are greatly enhanced around young dwarf stars in general, total water loss could be many times this value in some cases (Ribas et al. 2005; Ribas et al. 2010; Linsky et al. 2014).

Finally, an Earth-like planet could have CO₂ outgassing, plate tectonics and hence the potential for a carbonate–silicate weathering feedback (Walker et al. 1981). The CO₂ cycle on an initially anoxic planet without N₂ or Ar would be complex, because CO₂ condenses at relatively high temperatures (Lide 2000) but has low compressive strength in solid form (Clark & Mullin 1976). In the absence of ocean/interior heat transport processes, outgassed CO₂ could build up on the low instellation regions of a planet until the return flow of CO₂ glaciers became sufficient to transport it back to high instellation regions.

Setting aside the complexity of the full climate problem for future study, we can nonetheless demonstrate the potential for O₂ build-up in cases where CO₂ levels are such that the planet has an Earth-like global mean surface temperature. Figure 3 shows the variation of atmospheric temperature and H₂O molar concentration with atmospheric N₂ content calculated using the same methodology as in Wordsworth & Pierrehumbert (2013), for an Earth-like planet at 1 AU around a Sun-like star, assuming an N₂–CO₂–H₂O atmosphere with tropospheric H₂O relative humidity of 0.5. In each case, the CO₂ molar concentration has been chosen to yield close to T_s = 288 K in equilibrium. As can be seen, once the N₂ content drops below a few percent of that on present-day Earth, the high atmosphere becomes rich in H₂O, implying rapid photolysis and hence planetary oxidation. Hence we may conclude that even planets that are Earth-like in all respects except for the N₂ content of their atmospheres have the potential to build up O₂ abiotically until it is a major atmospheric constituent.

Zoom In Zoom Out Reset image size

Because O₂ can become the dominant gas in the atmosphere of a lifeless planet, alone it cannot be regarded as a robust biosignature. Our results do not necessarily rule out its utility in every case. However, they do demonstrate that the situation is considerably more complex than has previously been believed, with the likelihood of an abiotic O₂-rich atmosphere emerging a complicated function of a planet's accretion history, internal chemistry, atmospheric dynamics and orbital state. Investigation of the range of possibilities for terrestrial planets with variable N₂ and Ar inventories should be a rich area for future theoretical research that will help to expand our understanding of climate evolution mechanisms. Nonetheless, for a specific exoplanet, even detailed modeling might not lead to a definite conclusion given the inherent uncertainties in processes such as volatile delivery during formation.

Observationally, there may still be a way to distinguish the scenarios we discuss here, but only if a reliable way is developed to retrieve the ratio of O₂ to N₂ or Ar in an exoplanet's atmosphere. In principle this may be achieved by analyzing the planet's spectrally resolved phase curve (Selsis et al. 2011), in transit by measurement of the spectral Rayleigh scattering slope (Benneke & Seager 2012) in a clear-sky (i.e., aerosol-free) atmosphere, or possibly via spectroscopic observation of oxygen dimer features (Misra et al. 2014). More work will be required to assess the potential of these techniques to determine O₂/N₂ mixing ratios in realistic planetary atmospheres.

R.W. acknowledges support from the National Science Foundation and NASA's VPL program. This article benefited from discussions with F. Tian, R. de Kok, S. Rugheimer and D. Sasselov.