In putative cool, dry epochs, little to no surface water would have been available, and therefore water-rock reactions sourcing H2 would slow or cease. Although the majority of Fe oxidation was likely a result of crustal hydration (releasing H2 in a warm climate), slower Fe oxidation may have still occurred during periods of a cool, dry climate which would lack of surface aqueous systems. Early in Mars’ history, a lack of surface liquid water would expose reduced Fe(II) to the atmosphere, resulting in a loss of atmospheric oxidants to the ground; this scenario was previously considered by (25). Atmospheric oxidation of 2.7-Ga-old iron-nickel micrometeorites on Earth, potentially via high abundances of upper atmospheric oxygen (28–29) suggests early Mars may have undergone a similar loss of atmospheric oxidants to the ground. Chemical rates for iron oxidation by O2 have been measured at > 1000 K (36); however, the rate of iron oxidation by atmospheric oxidants is poorly constrained at temperatures relevant to the early Mars surface and are likely to be slow. In this work we vary the sink of oxygen from 106 cm− 2 s− 1 to 109 cm− 2 s− 1, with a lower bound representing a flux with negligible influence on the atmospheric chemistry and an upper bound within the range of oxidation weathering considered by (25). Under this climate, we assume crustal hydration (which we acknowledge may still occur in localized underground systems) is negligible during these cold periods.
We consider one case with no oxidant sinks as a basis for comparison. In this case, the photochemical results differ from the assumed warm scenarios as CO deposition to the ground goes to zero due to no aqueous processes. With no sink to the ground, CO mixing ratios are of a few percent in the 1 bar atmosphere and are near 10% in thinner atmospheres. CO is more abundant in thinner atmospheres due to less water availability. In all cases, we initialize water to the saturation vapor pressure, and in colder thinner atmospheres, less water is available to the recycling reaction: CO + OH → CO2 + H.
When ground sinks of oxygen are introduced, we find that the hydrogen escape rate increases. The loss of hydrogen to water in a 2:1 ratio seems to occur while the system has sufficient water to balance the oxygen sinks. However, once the oxidant sinks are increased beyond a rate that hydrogen diffusion can match (near ~ 109 cm− 2 s− 1), the hydrogen escape rate can no longer increase to balance the oxygen sinks. Instead, sinking oxygen begins to be pulled off of CO2 and the CO mixing ratio begins to rise. (25) comes to a similar conclusion that oxidant sinks result in a CO runaway in initially thick CO2-dominated atmospheres. Here, we improve the work with a more comprehensive model, including updated photochemical loss of carbon and coupling to a 1D climate model. Importantly, we find that the CO2 instability only occurs on timescales of > 107 years. This suggests that the response to obliquity variations (~ 105 years) and to transient H2 sources (~ 105 years) likely would have temporarily interrupted the evolution towards a CO atmosphere as the climate cooled. We find a CO-dominated atmosphere to be cooler (see Figure S1). The anharmonic mode coupling terms are small for CO because it is a diatomic molecule, resulting in fewer strong lines; CO has only an absorption band in the 5 micron region (which is far shortward of the main thermal distribution of a blackbody near 210 K) and a pure rotational band at longer wavelengths. Thus the main climate effect of CO would come from pressure-broadening of absorption by other gas species.
The behavior of 3D processes could have transported water to the upper atmosphere (such as periods of strong convection, variations in surface pressure, and obliquity changes), allowing rapid H escape. We investigate the effects of the water transport in only a 1D sense but find the oxidation of the surface to be faster than this oxidation of the atmosphere, maintaining relatively low O2 and large CO abundances (see Table S1).
We find that the mechanism of a CO runaway may also be self-limiting. The loss of atmospheric oxidants severely slows the recycling reaction CO + OH → CO2 + H, yielding a CO-dominated atmosphere in cases of larger H2 flux. The loss of a CO2-dominated atmosphere removes the main photochemical production pathway for O2, effectively placing an upper limit on the rate of O2 loss to oxidizing the surface. Larger sinks of O2 to the ground would therefore likely not be sustained on long timescales of > 108 years.
We hypothesize two possible mechanisms to destroy or prevent a CO-dominated atmosphere. First, on long timescales, the surface iron will become largely oxidized and thus sinks of atmospheric oxidants to the ground will slow. During the Amazonian era, the ground sink of oxygen lost to oxidize crustal iron would have been on average an order of magnitude less than the present-day atmospheric escape rate (17). Therefore, iron oxidation would minimally impact the atmospheric chemistry, and CO2 would be photochemically stable, as it is at present-day Mars. Second, a CO-runaway is most likely to occur in long-standing cool climates. Geologic events could have helped episodically warm a thinning atmosphere (on short timescales of ~ 105 years), thereby slowing the onset of the CO runaway.
We acknowledge that catalytic chemistry involving chlorine radicals is known to prevent a CO-runaway at Venus (e.g., 37). However, at present-day Mars, the HCl abundance is only a few ppb (38), which we predict would be insufficient to fully balance the CO runaway. The presence of present-day surface perchlorates (e.g., 21) implies the relevance of Cl- chemistry at early Mars and volcanism would likely have transiently increased HCl abundances to larger values; however, the HCl abundance at early Mars is poorly constrained and including Cl- chemistry in photochemical models would induce large error bars (spanning several orders of magnitude). For these reasons, we ignore catalytic chlorine chemistry, while also acknowledging its importance on the chemistry of other solar system worlds. Similarly, due to our focus on atmospheric chemistry and climate, we do not solve aqueous processes such as the potential for chlorates to oxidize ferrous iron at early Mars (39).