Special Articles

The Early Faint Sun Paradox: Organic Shielding of Ultraviolet-Labile Greenhouse Gases

See allHide authors and affiliations

Science  23 May 1997:
Vol. 276, Issue 5316, pp. 1217-1221
DOI: 10.1126/science.276.5316.1217


Atmospheric mixing ratios of ∼10−5 ± 1 for ammonia on the early Earth would have been sufficient, through the resulting greenhouse warming, to counteract the temperature effects of the faint early sun. One argument against such model atmospheres has been the short time scale for ammonia photodissociation by solar ultraviolet light. Here it is shown that ultraviolet absorption by steady-state amounts of high-altitude organic solids produced from methane photolysis may have shielded ammonia sufficiently that ammonia resupply rates were able to maintain surface temperatures above freezing.

The compositions and surface pressures of the early atmospheres of Earth and Mars after the end of the heavy bombardment are unknown. Models for Earth range from reducing atmospheres with surface pressures p s ∼ 1 bar (1) to p s ∼ 10 bar of CO2 (2). Atmospheres near a neutral oxidation state are currently favored for two reasons. First, a vast reservoir of CO2 is present in Earth’s sediments in the form of carbonates (p s ≃ 60 bar) and in the atmosphere of Venus (p s ≃ 90 bar). This carbon reservoir could have been in the early atmosphere partly as CO or CH4if some abundant reducing agent were present. However, metallic iron, one possible agent, is thought to have been segregated from the surface early and quickly during core formation (3), suggesting that carbon was present in the atmosphere primarily as CO2. The presence of other reduced gases such as NH3 also depends on the oxidation state of the upper mantle. If early continents were largely absent, the resulting inhibition of carbonate sedimentation may have led to dense CO2 atmospheres; Kasting (4) estimates that p s(CO2) was 0.1 to 10 bar 4 billion years ago (Ga). However, it is possible that the mantle may be more reducing than once thought (5), and geochemical arguments for early atmospheric composition are not yet conclusive. A second argument for an early atmosphere that was in a neutral oxidation state comes from photochemical kinetics, which suggests that atmospheric NH3 and CH4 would be photodissociated by solar ultraviolet (UV) radiation on short time scales (6-9). Here we argue instead that photodissociation of CH4 may have led to the production of a high-altitude organic aerosol that absorbed UV rays and allowed reducing gases to exist for much longer times. If this is correct, the second of the two arguments against reducing gases in the early atmosphere is no longer compelling, with implications for the origin of life and the resolution of the early faint sun paradox.

The origin of life requires sources of preexisting organic molecules, which are much more readily generated in reducing rather than in neutral oxidation-state atmospheres (10). With allowance for exogenous sources, reducing atmospheres give organic mixing ratios in the early oceans some three orders of magnitude higher than those for neutral atmospheres (11). Thus, models for Earth’s primitive atmosphere that postulate no significant abundance of reduced gases suffer the drawback that the origin of life is more difficult to understand.

The Early Faint Sun Dilemma

The equilibrium temperature of an airless, rapidly rotating planet is T e ≡ [SR 2(1 −A)/4a 2ɛσ]1/4, where σ is the Stefan-Boltzmann constant, ɛ the effective surface emissivity, A the wavelength-integrated Bond albedo,R the planet’s radius, a the planet’s distance from the sun, and S the solar constant at a. For Earth today, T e ≃ 255 K, and the difference ΔT ≃ 33 K between T e and the observed mean surface temperature T s is due to the mainly H2O-CO2 atmospheric greenhouse effect. Stars like the sun increase their luminosity through time as the hydrogen in the star’s core is converted to helium. Estimates of the change in S between 4 Ga and today are ≃0.25S 0 (12), with an error of about ±0.1S 0, where S 0is the present-day value. Holding A, ɛ, a, and the CO2 mixing ratio constant at their current values for Earth, and letting the H2O abundance be determined by its equilibrium vapor pressure, Earth’s T s < 0°C about 2 Ga (13-15). Allowance for an ice-albedo feedback implies a fully glaciated planet as recently as 1 Ga (16). These results contradict geological evidence for abundant, globally distributed liquid H2O before 3 Ga (13, 14, 17,18). The discrepancy is enhanced by evidence that before 2.5 Ga, Earth may have been considerably warmer than it is today (19). The implications are even more severe for Mars; yet there is strong evidence for liquid water on Mars 3.5 to 4.0 Ga (20).

This early faint sun paradox is only a paradox if we insist on holdingA, a, ɛ, and greenhouse warming constant over time. Were A = 0.3 at 4 Ga, as it is today,T e would have been 237 K (21); even if A had been 0 at 4 Ga, T e would still have been only 259 K. In models of an all-ocean early Earth with a 14-hour day, simulated temperatures rise only an additional 5.5 K (22). Likewise, at most a few degrees can be pried out of plausible long-term variations in infrared emissivity. Although some low-mass planets can chaotically migrate in heliocentric distancea over 4 billion years (23), there is no evidence for the steady increases in a for Earth by the ∼10% necessary to have compensated for the increase in S over its history. Another suggestion proposes that the early sun lost ∼10% of its mass, producing a solar flux at Earth before ∼4 Ga greater than its current value (24); however, there is as yet no clear evidence for such mass loss (25). Convolving the rapid rotation of Earth, an all-ocean planet, extremely low albedo, and a small solar mass loss, it is possible that early Earth reached 0°C but not much warmer temperatures. Such a conglomerate explanation seems strained. We therefore reconsider changes in greenhouse warming.

The Wien peak of Earth’s blackbody emission 4 Ga, as today, laid at 8 to 14 μm, a window in the combined absorption spectra of H2O and CO2 at pressures below a few bars and temperatures < 300 K. Because small quantities of NH3provide considerable opacity around 10 μm, Sagan and Mullen (13, 14) proposed a volume mixing ratio for NH3([NH3]) of ∼10−5, pressure-broadened in an ∼1-bar atmosphere, as the solution to the early faint sun dilemma for Earth and Mars. Subsequent calculations (6, 15) confirm that [NH3] ∼ 10−5 ± 1 would suffice. In comparison, background abundances in Earth’s current O2atmosphere are [NH3] ∼ 10−8(26). These NH3 abundances represent a steady state after loss processes, including oxidation, rainout, high-altitude photolysis, and reaction with CO2. Globally averaged, [NH3] ∼ 10−2 [CH4] today.

Although NH3 is soluble in water and subject to rapid rainout (27), there is, in the absence of photolysis, an equilibrium atmospheric abundance determined by oceanic mineral equilibria. Independently, a lower limit can be derived from the requirement that the deamination of aspartic acid, a key amino acid for the origin of life, be reversed. Bada and Miller (28) thereby found [NH3] for early Earth to be 10−7 at 0°C, 3 × 10−5 at 25°C, and 3 × 10−4 at 50°C, consistent with the 10−5 ± 1 derived from the greenhouse requirement. Other estimates lie in a similar range (29).

All organisms on Earth are able to metabolize NH4 + (30). Few organisms can process N2 directly, and ubiquitous metabolic facility with NH4 + may be a remnant from earlier times when atmospheric NH3 was readily available (31). These chemical and biological arguments independently point to NH3 as a minor constituent of the primitive atmosphere.

NH3 Photolysis and CO2Greenhouses

Photochemical kinetics demonstrates that [NH3] ∼ 10−5 would be photodissociated by the solar UV continuum at wavelengths λ < 2300 Å and irreversibly converted into N2 within decades (6, 8, 9). Likewise, [CH4] ∼ 10−5 would be photolyzed within centuries. These time scales are so short that hypotheses about resupply (13, 14) seem wholly without merit; an NH3 outgassing rate ∼5 × 1015 g year−1 would be required to keep pace with photolysis, which implies ∼103 bar of N2 built up over ∼1 billion years, contradicting current atmospheric and sedimentary rock inventories.

Owen et al. (8) proposed that massive amounts of CO2 might close the 8- to 14-μm window and provide the missing greenhouse effect. Numerous calculations have since shown that a few bars or less of CO2 might suffice to raise the surface temperature of Earth 4 Ga to several tens of degrees Celsius (15, 32). But Rye et al. (33) argue from the absence of siderite in paleosols that [CO2] between 2.75 and 2.2 Ga was ≤10−1.4 bar, implying that greenhouse gases other than CO2 contributed to greenhouse warming in the late Archean and early Proterozoic eons. For Mars, there is extensive evidence for ancient outflow channels and valley networks (20), some evidence for lakes (34), and even an argument for oceans (35) 3 to 4 Ga. The general opinion has been that this evidence requires a massive early greenhouse effect on Mars (14, 15, 36), but the higher early subsurface thermal gradient eases the greenhouse demands (37). Massive CO2 greenhouse models face challenges from CO2condensation (38) and the difficulty in finding sufficient corresponding martian carbonate sediments (39).

Ultraviolet Shielding

Given doubts about massive CO2 atmospheres for Earth and Mars, we reexamine models of comparatively thin atmospheres in which minor constituents provide the needed infrared opacity. We ask whether, with no ad hoc assumptions, there is a way to shield NH3 and other UV-labile gases so that demands on the resupply rate are reasonable. Methane can be dissociated only by UV photons of λ < 1450 Å (the dissociation is therefore entirely dominated by solar Lyman α radiation), whereas NH3photolysis is driven by UV at wavelengths as long as 2300 Å. This longer wavelength UV, which in total energy flux exceeds that of λ < 1450 Å UV by more than two orders of magnitude, penetrates deeper into the atmosphere. The result is that whereas CH4 photolysis is expected to occur high in the atmosphere (40, 41), photolysis of NH3 should occur at a lower altitude, with NH3 photolysis products spatially separated from those of CH4 (6).

Photochemical modeling indicates that NH3 photolysis, for [NH3] ∼ 10−5 ± 1, peaks at altitudes of 25 to 35 km, depending on [NH3] and the assumed eddy diffusion coefficient (6). Zahnle (42) has modeled the altitude at which [CH4] photolysis and polymerization should occur as a function of [CH4], [CO2], UV flux, and eddy diffusion profile. For contemporary values of the eddy diffusion profile and [CO2], he finds the altitude of CH4 polymerization ranging from 40 to 75 km (43) as [CH4] varies from 10−5 to 6 × 10−4 (corresponding to CH4 fluxes into the atmosphere ranging from 5 × 1010 to ∼4 × 1011 cm−2 s−1). In all cases, he takes [H2O] at the base of the stratosphere to be 4 × 10−6; that is, CH4 polymerization is not prevented by OH generated from UV dissociation of H2O (44). However, a transition from oxidation of CH4 to polymerization occurs when the local C/O ratio passes through unity, at which point large percentages of CH4 are polymerized. The CH4/CO2ratio is therefore important.

Organic solids from atmospheric shocks (as well as interplanetary dust particles) contribute to a high-altitude layer of organic aerosols, depending on atmospheric composition (11). If CH4 photolysis and other processes produce an organic haze, which in turn provides UV shielding to greater depths (the C2 hydrocarbons absorb at considerably longer wavelengths than CH4), NH3 will lie below this screen and be substantially shielded from photolysis. Experimentally, N2-CH4 gas mixtures with [CH4] ∼ 10−3 generate upon irradiation complex organic solids with strong UV absorption in the 2000 Å region (45). Destruction of NH3 by reaction with OH (9) should also be greatly reduced, because photolysis of H2O from photons with λ < 2400 Å will be suppressed. Submicrometer aerosols should settle out of the stratosphere on time scalest ≃ 0.5 to 3 years (46), fairly independent of the atmospheric pressure on early Earth;t ≃ 1 year applies to the present Earth (p s = 1 bar, gravitational accelerationg = 980 cm s−2) and Venus (p s = 90 bar, g = 890 cm s−2).

To calculate the prolongation by shielding of the ∼10-year photodissociation time scales of unshielded NH3, we must evaluate two photodestruction rates. The photodestruction rate in the absence of any shielding is J = ∫I λσλ dλ, whereI λ is the solar flux at the top of the atmosphere (47) and σλ is the NH3 absorption cross section (9). If a high-altitude UV shield of optical depth τλ is present, this rate becomes J′ = ∫I λ e τλ sec θσλ dλ, where θ is the solar zenith angle; we take θ = 45°, the mean angle. These integrals effectively extend only from 1100 to 2300 Å (Fig. 1), because the solar flux is negligible below 1100 Å and NH3 absorption is negligible above 2300 Å. The ratio J/J′ therefore gives the enhancement of the lifetime of atmospheric NH3 in the presence of a UV shield of optical depth τλ. Because σλ is increasing steeply and the Planck function is decreasing steeply at 2000 Å, these integrals are given approximately by their values near 2000 Å. Before performing the full numerical integrations, we first examine such an approximation.

Figure 1

Solar UV flux at Earth (solid line) and the absorption cross section of NH3 (dashed line) as a function of wavelength. With the exception of the Lyman α spike at 1216 Å, solar UV flux drops quickly shortward of 2000 Å, whereas NH3 absorption falls rapidly longward of 2000 Å.

All organic solids strongly absorb at λ < 2300 Å. The optical depth is given by τλ = φγ tαλ/4πρR 2, where φγ is the production flux (mass per time) of UV-shielding organic solids, R is the planet’s radius, ρ is the bulk density of the organic solids, αλ = 4πk/λ is the linear absorption coefficient of the organic solids, and k is the imaginary part of the complex refractive index (Table 1). We choose a conservative value of k(2000 Å) ≃ 0.19. The higher the proportion of elemental carbon, the higher k will be. We adopt ρ ≃ 1.4 to 1.6 g cm−3, the range for the organic solids produced at low pressures from the radiation chemistry of 10:1 N2-CH4 gas mixtures (48).

Table 1

Values of the imaginary part of the complex refractive index k at 2000 Å for candidate organic solids that could serve as a UV shield.

View this table:

Syntheses of organics in CH4-N2 models of Titan’s atmosphere imply a production efficiency of 1.2 × 10−12 g erg−1 at λ < 1550 Å (49). This value is within a factor of 2 of the photodissociation efficiency of CH4 in pure CH4 atmospheres at λ ≃ 1295 and 1470 Å (50). We calculate organic production efficiencies in UV-irradiated CH4-NH3 atmospheres (51) to be a factor of ∼5 below those just cited for CH4-N2 atmospheres.

Most of the products of charged-particle irradiation of CH4-N2 atmospheres are in the form of organic heteropolymer (52). A lower limit on the fractionf of the irradiation products that are organic solids is taken to be 0.1 (52). We assume a similar range for the products of CH4-N2 photolysis. We estimate the net flux of UV energy below 1500 Å on Earth 4.0 Ga to have been 1 × 1027 erg year−1 (11, 53), three times the current value. The production of high-altitude organic aerosols from CH4 photolysis in the terrestrial atmosphere 4 Ga was then φγ ≃ (1.2 × 1015)f g year−1. Other sources of high-altitude organic aerosols (11) amount to ≲1013 f g year−1.

This calculation assumes the production of organic aerosols to have been UV-limited. The contemporary carbon outgassing rate (mainly as CO2) from mid-ocean ridges is 1 × 1013to 1 × 1014 g year−1; estimates (54) of outgassing rates 3 Ga range from 4 × 1013 to 6 × 1014 g year−1(or ∼1011 cm−2 s−1). The outgassing rate 4 Ga must have been greater still but with an unknown CH4/CO2 ratio. It seems possible, even without exogenous input (55), that the carbon outgassing source would have been able to keep pace with the organic solid production sink, provided CH4/CO2 ≳ 1. Integrated over 1 billion years, the organic production would have been of the same order as the total estimated terrestrial inventory of carbon in the mantle and crust (56).

NH3 Lifetimes, Diffusion Flux, and Resupply

Choosing k(2000 Å) = 0.19, appropriate for kerogen, with ρ = 1.6 g cm−3, t = 1 year, and f = 0.5, we find αλ = 1.2 × 105 cm−1 and τ ≃ 8.8, yielding an attenuation of NH3 photolysis by a factor exp(−τ sec θ) = 3.9 × 10−6. The lifetime of [NH3] ∼ 10−5, instead of 10 years, would then be 2.5 × 106 years. The rate of NH3photodestruction would be ∼2 × 1010 g year−1 in this case or, integrated over 1 billion years, about 2 × 1019 g, or 0.5% the current atmospheric N2 abundance. By comparison, the contemporary biogenic value is ∼1013 to 1014 g NH3year−1 (9, 31, 57). In CH4-rich reducing atmospheres, NH3 is also a degradation product of HCN; steady-state production is estimated at ∼1012 ± 1g NH3 year−1 (58).

Models with k(2000 Å) = 0.22, appropriate for N2-CH4 irradiation residues (Titan tholin), with ρ = 1.4 g cm−3, t = 1 year, and f = 0.5 give τ ≃ 11.7, or a [NH3] ∼ 10−5 lifetime of 1.5 × 108 years. Averaging the values for kerogen and Titan tholin, the lifetime of a [NH3] ∼ 10−5 primeval atmosphere as a function of f and sedimentation time t fork(2000 Å) ≃ 0.21 and ρ = 1.5 g cm−3shows that effective shielding of NH3 by organic solids fails only for low t and low f (Table2).

Table 2

Lifetimes for atmospheres with [NH3] ∼ 10−5. The sedimentation times are from Oberbecket al. (46). E indicates a value >4.5 billion years.

View this table:

The full integrations of J/J′ for kerogen optical constants, ρ = 1.6 g cm−3,t = 1 year, and f = 0.5 giveJ/J′ = 1.0 × 105, or lifetimes for [NH3] ∼ 10−5 only a factor of 2.5 less than those found by the 2000 Å approximation (Fig. 2). [The net solar UV flux 4 Ga for λ < 2300 Å was about the same as it is today (11, 53).] For Titan tholin optical constants and ρ = 1.4 g cm−3, the result isJ/J′ = 9.6 × 106, or a lifetime of ∼108 years. As a consistency check, we note that for many of the organic solids (Table 1), τ(5500 Å)/τ(2000 Å) ∼ 0.1 or less, so that even for τ(2000 Å) ∼ 10, there is still enough sunlight reaching Earth’s surface to drive a significant net greenhouse effect.

Figure 2

Solar UV flux with no attenuation (heavy line), compared with the flux at Earth’s surface in the presence of a high-altitude organic solid formed by reactions of UV light with CH4. The dashed line shows the UV flux if the high-altitude haze is assigned kerogen optical constants. The dotted line shows the same, but with optical constants appropriate to Titan tholin.

However, atmospheres in which NH3 lifetimes are sufficiently prolonged will violate the assumption that NH3photolysis occurs below the organic aerosol shield. For example, forf = 0.5 and 1-year sedimentation times (Table 2), shielding lengthens NH3 photolysis lifetimes by a factor ∼ 106. In this case, the time scale for NH3 to reach the altitude of CH4 polymerization would be shorter than the photolysis lifetime, and the NH3would no longer be shielded. Then the required NH3 resupply flux φ will be given by its flux, dominated by eddy diffusion (59), to the CH4 polymerization altitude. Contemporary eddy diffusion constants K vary from ∼103 cm2 s−1 just above the tropopause to ∼106 cm2 s−1 at an altitude of 80 km (60). If we choose K = 105 cm2 s−1 for the early atmosphere (6), and an atmospheric scale height H= 8.4 km, an upper limit (61) for the NH3 eddy diffusion flux is φe ≈ (K/H)[NH3]n(z) ≈ 2 × 109 cm−2 s−1 or ∼1013 g year−1 at z = 70 km; here n(z) is the contemporary atmospheric number density at altitude z. {Lower polymerization altitudes would give higher φe due to largern(z) [n(z = 40 km) ∼ 40n(z = 70 km)], although this effect is mitigated by lower values of K at lowern(z).} In this case, [NH3] ∼ 10−5 has a lifetime of ∼[NH3]M a/φ ∼ 5000 years (where M a = 5 × 1021 g is the mass of the contemporary atmosphere), so that the current inventory of N2 would be generated by NH3 photolysis in ∼5 × 108 years. Given resupply, half-billion-year time scales for the duration of [NH3] ∼ 10−5 atmospheres appear possible.

Would CH4 have been irreversibly converted to organic solids and NH3 to N2 and organic solids on the early Earth, or are there mechanisms for recycling? Tholins are stable and require temperatures of 600° to 1000°C for half their mass to be vaporized (62). If plate tectonics were in operation 4 Ga, with subduction of the crust down to hundreds of kilometers depth and a geothermal gradient substantially steeper than that which exists today, tholins may have been reprocessed into NH3, CH4, and other simple organic gases. Recycling of N2 into NH3 may occur by means of electrical discharges and NO (30) and TiO2 photochemical catalysis in titanium-rich deserts (63). These processes reduce the need for outgassing or exogenous resupply. Because methanogens were among the earliest microorganisms, and life seems to have been widespread by 3.5 and possibly by 3.85 Ga (18), there is no need to seek nonbiological sources of CH4 (or NH3) in that period or later. Ultraviolet optical depths above 5 to 10 on Earth would have played an important role in shielding early organisms from UV damage (64).

A point of comparison for this model is provided by Saturn’s moon Titan. The UV and radiation processing of Titan’s N2-CH4 atmosphere produces an aerosol with average particle radii ∼0.25 μm and an optical depth of about 7 between 2000 and 3000 Å (65), about the same as the optical depths calculated here for CH4 photolysis in the early terrestrial atmosphere.

There are clearly uncertainties in many of the input parameters in our calculations. Nevertheless, were the atmosphere of the early Earth reducing, our results suggest that it would have been self-shielding against UV photodissociation. Were this atmosphere instead rich in N2 with minor CO2 and CH4components, self-shielding may also have been possible, but only if a CH4/CO2 ratio ≳1 could have been maintained. We wish only to suggest that a self-consistent solution to the early faint sun dilemma (66), without invoking massive CO2 atmospheres, appears to be viable.


View Abstract

Stay Connected to Science

Navigate This Article