|Home | About | Journals | Submit | Contact Us | Français|
The release and oxidation of ferrous iron during aqueous alteration of the mineral olivine is known to reduce aqueous solutions to such extent that molecular hydrogen, H2, forms. H2 is an efficient energy carrier and is considered basal to the deep subsurface biosphere. Knowledge of the potential for H2 generation is therefore vital to understanding the deep biosphere on Earth and on extraterrestrial bodies.
Here, we provide a review of factors that may reduce the potential for H2 generation with a focus on systems in the core temperature region for thermophilic to hyperthermophilic microbial life. We show that aqueous sulfate may inhibit the formation of H2, whereas redox-sensitive compounds of carbon and nitrogen are unlikely to have significant effect at low temperatures. In addition, we suggest that the rate of H2 generation is proportional to the dissolution rate of olivine and, hence, limited by factors such as reactive surface areas and the access of water to fresh surfaces. We furthermore suggest that the availability of water and pore/fracture space are the most important factors that limit the generation of H2. Our study implies that, because of large heat flows, abundant olivine-bearing rocks, large thermodynamic gradients, and reduced atmospheres, young Earth and Mars probably offered abundant systems where microbial life could possibly have emerged. Key Words: Serpentinization—Olivine—Hydrogen—Deep biosphere—Water—Mars. Astrobiology 11, 711–724.
Hydrous alteration of the mineral olivine (Mg2(1−x)Fe2xSiO4) is believed to be a key subsurface source for abiotic molecular hydrogen (H2) generation (e.g., Sleep et al., 2004; Seyfried et al., 2007; Klein et al., 2009; McCollom and Bach, 2009). Olivine is the main mineral component in ultramafic rocks such as peridotite and dunite, which are abundant in the mantle, in seafloor crustal accumulates, and in ophiolites which represent sections of oceanic lithosphere obducted onto the continental crust (e.g., Rossman et al., 1989; Albrektsen et al., 1991; Dick et al., 2000; Nicolas and Boudier, 2000; Iyer et al., 2008). Moreover, olivine is abundant in the martian crust in ultramafic accumulates and in picritic ultramafic lavas (Hoefen et al., 2003; Christensen et al., 2005; Mustard et al., 2005). As olivine alters and releases ferrous iron (Fe2+), molecular oxygen (O2) acts as the main electron acceptor in the generation of ferric iron (Fe3+) and is consumed. This removal of molecular oxygen (O2) from groundwaters ultimately leads to reduction of water into H2 (e.g., Neal and Stanger, 1983; Oze and Sharma, 2007; Hellevang, 2008; McCollom and Bach, 2009).
Molecular hydrogen is observed to seep out in gas mixtures in ophiolites (e.g., Neal and Stanger, 1983; Abrajano et al., 1988, 1990), in hydrothermal vent fluids hosted in ultramafic rocks at and near slow-spreading mid-ocean ridges (e.g., Kelley et al., 2001, 2005; Charlou et al., 2002), and in continental groundwater and hydrothermal fluids (Stevens and McKinley, 1995; Spear et al., 2005). H2 forms at these settings by alteration reactions where mafic and ultramafic minerals dissolve and are replaced by hydrated mineral assemblages (Moody, 1976; Kelley et al., 2002, 2004; Früh-Green et al., 2004; Bach et al., 2006; Schulte et al., 2006; Oze and Sharma, 2007; Bach and Früh-Green, 2010).
Knowledge of the alteration of olivine and its potential to generate H2 is important to the understanding of topics such as the emergence and evolution of microbial life (Sleep et al., 2004; Russell, 2007; Milner-White and Russell, 2008; Sleep and Bird, 2008; Russell et al., 2010), the use of H2 and H2-derived species as possible tracers for ongoing water-rock interactions in environments such as the martian surface (e.g., Atreya et al., 2007; Mumma et al., 2009; Parnell et al., 2010), and the deep subsurface biosphere driven by hydrogen or hydrogen-derived species (e.g., Gold, 1992; Stevens and McKinley, 1995; Pedersen, 1997; Chapelle et al., 2002; Kelley et al., 2004; Takai et al., 2004; Nealson et al., 2005; Spear et al., 2005). Microbes in the deep biosphere encompass a wide range of energy metabolisms (e.g., Kaplan et al., 1963; Claypool and Kaplan, 1974; Teske et al., 2003; Takai et al., 2004; Schippers et al., 2005; Schippers and Neretin, 2006; Wortmann et al., 2007; Adhikari and Kallmeyer, 2010), in which H2 is an integral part (e.g., Morita, 2000; Spear et al., 2005). The basal nature of H2 in these systems suggests that, to some extent, the deep biosphere is dependent on the generation rate of H2 or H2-derived species (e.g., Hellevang, 2008).
It has been recognized that the potential for H2 generation is dependent on the partitioning of ferrous iron between mineral phases, the abundance of unaltered olivine, and temperature (e.g., Sleep et al., 2004; Seyfried et al., 2007; Klein et al., 2009; McCollom and Bach, 2009). The aim of our study was to evaluate additional factors that may potentially reduce the rate and amount of H2 production in subsurface systems. Because a large part of seafloor and ophiolite alteration occurs in low-temperature systems, and the upper temperatures of microbial growth have been recorded to 121–122°C (Kashefi and Lovley, 2003; Takai et al., 2008), we limited this review to relatively cold systems. For thermodynamic calculations, we used 80°C, which is a core temperature for thermophilic to hyperthermophilic microorganisms (Brock, 1985; Kashefi and Lovley, 2003; Roussel et al., 2008; Torsvik and Øvreås, 2008). The factors that are covered are related to the effect of aqueous redox reactions in the nitrogen, carbon, and sulfur systems on mineral reaction rates and hydrodynamic conditions; and a brief section on the constraint of H2 formation on the availability of water is given. Finally, the implication of inorganic hydrogen generation on the emergence of life on Earth and on the potential for hydrogen production and life on extraterrestrial planets is discussed. The results are organized by presenting factors that limit H2 generation, together with numerical simulations.
All thermodynamic calculations were performed by using the geochemical code PHREEQC-v2 (Parkhurst and Appelo, 1999). The standard state employed was unit activity for pure minerals and H2O at any temperature and pressure. For aqueous species other than H2O, the standard state was unit activity of the species in a hypothetical 1 molal solution referenced to infinite dilution at any temperature and pressure. For gases, the standard state was unit fugacity of a hypothetical ideal gas at 1 bar of pressure.
We used the thermodynamic data for the minerals and aqueous species listed in the phreeqc.dat database with the following modifications: thermodynamic data for the minerals brucite (Mg(OH)2), magnetite (Fe3O4), fayalite (Fe2SiO4), and forsterite (Mg2SiO4), were copied from the phreeqc llnl.dat database file, as was the aqueous redox speciation of Fe. Activities of charged aqueous species were calculated according to the WATEQ Debye-Hückel equation (Truesdell and Jones, 1974). The olivine composition used in the calculations is Fo90, which represents a solid-solution mix between the forsterite (90%) and fayalite (10%) end members. The numerical simulations were done at 80°C and 1 bar pressure. For the sulfur and carbon systems, an initial gas phase was defined to consist of molecular nitrogen (N2) and O2, corresponding to the present-day atmosphere. The initial aqueous solution had a NaCl content that closely corresponded to present-day seawater (Table 1).
The potential for H2 generation is dependent on the amount of ferrous iron supplied to the aqueous solution from olivine, the subsequent oxidation of the ferrous iron to ferric iron, and the precipitation of ferric iron into minerals such as magnetite (Fe3O4). The maximum potential for H2 generation is given if magnetite is the only secondary mineral (e.g., Seyfried et al., 2007; Hellevang, 2008; McCollom and Bach, 2009). However, the potential is reduced if ferrous iron is removed from the aqueous solution without oxidation. This happens when ferrous iron is incorporated into other secondary minerals like solid solutions of ferrous brucite (Mg1−xFex(OH)2) and Fe-serpentine (MgFe(II)Fe(III))3Si2O5(OH)4 (Seyfried et al., 2007). The extent of iron substitution and the effect on H2 generation was treated in detail by Seyfried et al. (2007) and will not be repeated here. We examine below three additional factors that associate this process for the potential for H2 generation: (a) the dissolution rate of olivine and hence the supply rate of ferrous iron to the aqueous solutions; (b) the presence of other aqueous redox couples that directly or indirectly affects the aqueous activities of O2 and hence H2; and (c) pore space and the availability of water.
The amount of ferrous iron supplied to the water from olivine within a given time interval is given by the amount of the mineral that dissolves multiplied with the stoichiometric moles of iron per mole olivine. For a typical seafloor olivine (Fo90=Mg1.8Fe0.2SiO4), this corresponds to
where n denotes moles, t time, and v the stoichiometric coefficient (v=0.2 for Fo90). Data of far-from-equilibrium olivine dissolution rates are available from several reports for a range of conditions (e.g., Grandstaff, 1978; Pokrovsky and Schott, 2000; Rosso and Rimstidt, 2000; Oelkers, 2001; Golubev et al., 2005). The laboratory experiments suggest that the dissolution rate has a 0.5 order dependence on proton activity at acidic conditions (Pokrovsky and Schott, 2000; Rosso and Rimstidt, 2000; Oelkers, 2001), whereas the rates appear to become pH independent at pH>9 (Pokrovsky and Schott, 2000). The dissolution rate can therefore be written as
where is the average reactive surface area expressed in square meters, is the H+ activity, and the 25°C dissolution rate coefficients k+,1 and k+,2 can be expressed as (Pokrovsky and Schott, 2000)
where  denotes concentration, >i are surface complexes, and Kex are the apparent equilibrium constants for the surface adsorption and exchange reactions, respectively. The implications for olivine dissolution rates and H2 generation are that any adsorbents that compete for olivine surface sites may affect the rates. The temperature dependence of the dissolution rate has been found to follow an apparent activation energy of 63.8kJ/mol, which at low temperatures corresponds to an increase of approximately 1 order of magnitude per 30°C increase (Oelkers, 2001).
A simplified model for the temporal evolution of H2 formation in seafloor hydrodynamic systems can be represented by a moving water mass as a closed-system package that is not mixed with other aqueous solutions but reacted with olivine according to a kinetic rate law. The time needed for waters to alter olivine and generate H2 in a hydrothermal system can then be estimated. We simulated such a package represented by a saline aqueous solution in contact with Fo90 at 80°C. The kinetic coefficients were estimated from Oelkers (2001) and Pokrovsky and Schott (2000). Ten moles of Fo90 were initially defined to react with 1kg of solution. No dependence was made between the reactive surface area and the mass of Fo90 present, and the rate was hence independent of the mass until all Fo90 was dissolved and the reaction stopped. As the average reactive surface area of olivine is likely to vary by orders of magnitude because of differences in grain size and degree of exposure (pore space versus fracture), we varied from 0.1 to 10m2.
The temporal evolution of H2 generation for various average reactive surface areas is shown in Fig. 1. For a solution with a reactive surface area of 10 square meters per kilogram of water (m2/kgw), H2 at a millimolar concentration and higher was reached after only 6 years of Fo90 alteration. At 1m2/kgw, the time needed to reduce the solution and form H2 increased to about 60 years. Because the rate of Fo90 dissolution was nearly constant during the simulated times, the time t needed to generate millimolar concentration of H2 followed , that is, increased inversely proportional to the reactive surface area. However, as key factors such as average reactive surface area and temperature are likely to vary with time as a water package travels in a circulating system, such simple linear predictions are not applicable to natural systems.
Another factor that may affect the dissolution rate of olivine is surface passivation by secondary coatings or modifications of the surface morphology. From laboratory experiments, Béarat et al. (2006) observed that a layer of secondary minerals formed on dissolving olivine and explained this formation by a combination of non-stoichiometric dissolution leaving a silica-rich leached layer and the precipitation of amorphous silica. The progress of olivine alteration has been explained partly by exfoliation and partly by diffusive transport of ions through the surface layer (Kim et al., 2005; Béarat et al., 2006). Welch and Banfield (2002) found that the reactivity of olivine in acid solution in the presence of Acidithiobacillus ferrooxidans was passivated by catalyzed surface oxidation and adsorption of Fe3+. At present, the extent of passivation is still not well understood. We thus conclude that, in natural systems where secondary coatings are formed and affected by microbial activities and passivating layers grow in a confined environment without exfoliation, diffusion-controlled dissolution with progressively lower release rates with time may reduce the potential for H2 generation.
Nitrogen, sulfur, and carbon occur in a range of redox states in natural waters. These compounds may affect the degree aqueous solutions can be reduced and, hence, the effect on the potential for H2 formation.
Nitrogen occurs from the most-reduced cyanide (CN−) and ammonium (), with N valence states of −5 and −3, respectively, to the oxidized nitrate () where nitrogen is in a N(+5) valence state. Molecular nitrogen (N2) is in an intermediate redox state with N(+0). In redox reaction sequences, is second after O2 and followed by oxidized manganese and iron. This means that the thermodynamic potential for to oxidize other compounds such as ferrous iron is second after O2. The present-day deep seawater has a N2 concentration of ca. 16ppm (Craig et al., 1967), whereas dissolved is below 2.5ppm (Johnson et al., 1989). At low temperatures and without catalysts, the formation of N2 from the reduction of and (nitrite) and the oxidation of is irreversible because of the strong triple-bonded nitrogen atoms in N2. However, in the presence of catalysts such as metallic Fe and Ni and Fe-Ni alloys, and at reducing conditions, nitrogen has been shown to be further reduced to ammonium even down to low temperatures (Smirnov et al., 2008). The reduction of to N2 can be written in two steps through the unstable intermediate nitrite:
where the double arrow indicates equilibrium and the right arrow indicates an irreversible reaction in the forward direction. Assuming equilibrium between and N2, the aqueous O2 activity in a system containing and N2 can be found from
where K(5) is the equilibrium constant for reduction of to N2. As oxidation of ferrous iron consumes O2, is irreversibly decomposed to N2 and O2, and is correspondingly consumed according to the equilibrium in Reaction 5. Because present-day deep seawater contains less than 2.5ppm , we conclude that, even at the extreme case where equilibrium is assumed, does not interfere with H2 generation.
As for nitrogen, sulfur occurs in a range of redox states from the oxidized sulfate (SO42−) with S(+6) to the reduced H2S or HS− with S(−2). The reduction of SO42− to HS− can be written in a series of aqueous equilibrium reactions according to
If equilibrium between SO42− and all intermediates in the reduction series (7) are assumed, the direct reduction of SO42− to HS− can be written as
Then from the law of mass action, the activity of O2 is
where K(8) is the equilibrium constant for Reaction 8. This expression suggests that, given thermodynamic equilibrium in the aqueous sulfur system, the activity of O2 is constrained by the activity ratio of SO42− over HS−. As reduced conditions favor the stability and growth of sulfide minerals and elemental sulfur, the total amount of sulfur in the aqueous solution will be reduced by the mineral growth, and the SO42− activity will be correspondingly reduced.
To estimate the effect of SO42− on the potential for H2 generation, we performed numerical simulations of Fo90 alteration with initial SO42− concentrations of 0.029mol/kgw, which corresponds to that of seawater (Nordstrom et al., 1979), and compared to simulations without initial SO42− (Fig. 2). The mineral pyrite (FeS2) was selected as proxy to illustrate the effect of secondary sulfide mineral growth on the aqueous sulfur activity. Examples of other sulfide-containing minerals that are likely to form in these systems and would have a similar effect are pyrrhotite (Fe1−xS) and mackinawite ((Fe,Ni)1+xS). The simulations show that, when Fo90 reacts in a sulfate-containing solution, SO42− is reduced to HS− and H2S as O2 is consumed by the oxidation of the ferrous iron (Fig. 2a). As the reaction proceeds, pyrite precipitates, which reduces the amount of sulfur in the water-gas system (Fig. 2b). Eventually, after ca. 0.08mol of Fo90 have dissolved (11.76g in 82mL of water), SO42− drops to <10−15mol/kgw, and H2 is produced to high concentrations (Fig. 2c, 2d). A comparison between the effect of Fo90 alteration on the aqueous O2 activity and the amount of gas-phase H2 with and without sulfur is shown in Fig. 3. The simulations suggest that the O2 activity drops rapidly to <10−67 (Fig. 3a), and H2 is correspondingly produced at percent levels (Fig. 3b) in a system without oxidized sulfur compounds. On the other hand, in a system that contains seawater level of SO42−, the O2 activity remains above 10−64 (Fig. 3a), and the H2 in the gas phase is only at part-per-million levels until 0.08mol of Fo90 is reacted (see Fig. 2c). At high Fo90/water ratios, however, the time required to dissolve 0.08mol/kgw of Fo90 is likely short, and the effect of SO42− is hence limited.
Natural waters contain dissolved inorganic and organic carbon, and particulate inorganic and organic carbon, with the carbon valence states from the most oxidized CO2 species with C(+4) to reduced species such as methane (CH4) with C(−4). Ocean water in contact with an atmosphere exchanges CO2 and approaches thermodynamic equilibrium with the gas. The dissolved CO2 reacts with water and forms carbonic acid, H2CO3, which dissociates to form bicarbonate and carbonate ions:
At progressively more reducing conditions, the C(+4) carbon in dissolved CO2 is thermodynamically favored to be reduced in a series of steps to CH4
Assuming equilibrium among all intermediates in Reaction 11 or the direct reaction of CO2 with water to form CH4, the redox equilibrium between aqueous CO2 and CH4 can be expressed by
The O2 activity is then fixed at equilibrium between CO2,aq and CH4,aq by
where K(12) is the equilibrium constant for Reaction 12. In closed systems, a constraint on the upper aqueous CO2 activity is given by carbonate minerals as they are stabilized at increasing CO2 contents and precipitate out (Griffith and Shock, 1995).
We simulated the effect of carbon on the potential for H2 generation at various initial amounts of dissolved CO2 and conditions as given in Table 1. Figure 4 shows the aqueous activity of H2 and CH4 as Fo90 is dissolved in the aqueous solution. Initial aqueous inorganic carbon concentrations were varied from 0.0012 to 0.12M. At the lowest carbon content, the aqueous solution is rapidly reduced and forms significant amounts of H2 (Fig. 4). As the carbon content increases by 1 order of magnitude to 0.012M, more Fo90 is required to react before high levels of H2 are reached. Finally, upon further increasing carbon by 1 order of magnitude, the H2 activity does not reach values higher than 10−9 for the added Fo90. In simulations shown in Fig. 4, the amount of carbon is constant, but the CO2 pressures vary. As the aqueous solutions are reduced, aqueous CO2 may be inorganically or biologically converted to CH4, and the CO2 pressure drops. This allows the O2 activity, expressed in Eq. 13, to drop sufficiently to allow significant H2 formation. In open systems with constant CO2 pressure, the CO2 aqueous activity is fixed by the corresponding CO2 fugacity, and the potential for H2 generation is likely to be more limited.
The alteration of ultramafic or mafic rocks and minerals leads to the consumption of water into less dense hydrous minerals. The consequence of this is volume expansion, reduction of the pore space, and a consumption of pore-space water. The closed-system isochemical alteration of olivine corresponds to the type 1 rock-dominated alteration as defined by Beard et al. (2009), where pre-existing voids and fractures are the sources of water and the alteration is observed to be limited.
The first factor that limits the olivine alteration in a closed system is the amount of water available. For a given amount of water, a maximum alteration potential is given from the stoichiometric amount of water incorporated into the hydrous phases. For example, the alkaline isochemical alteration of Fo90 into pure Mg-serpentine and Mg-brucite, magnetite, and H2 can be written as
Each mole of Fo90 altered produces 0.5mol of serpentine, 0.3mol of brucite, and hence consumes 41/30mol of water. This means that approximately 24.63g of water is consumed for every 147g of Fo90 altered, or ca. 24.63mL water is consumed for every 43.49mL Fo90 altered. Any lower water content will only lead to partial alteration, until more water is gained access to the fresh olivine.
The second factor is the volume expansion of the solid phases that reduces the pore volume of the rock. The volume changes following olivine hydration include both the volume changes of the pore space and the aqueous phase. The total change in mineral volume ΔVmin can be estimated by the difference in reduced volume following olivine dissolution and the added volume of the secondary phases:
where ξ is the reaction progress given as moles of olivine released to the aqueous solution, V is volume, and subscripts i and ol denote secondary phases and olivine, respectively. Because of the lower densities of the secondary phases compared to olivine, the mineral volume changes are always positive. For a water-saturated closed system, the volume change of the aqueous solution corresponds to the pore-volume change, leading to changes in the specific volume of the aqueous phase and corresponding pressure changes. The specific volume is given by
where V is the specific volume, and V and n denote volume and moles of the aqueous phase, respectively. The pressure can be estimated by using a proper equation of state for water such as defined by Mao and Duan (2008). Equations 15–16 show that if the volume of aqueous solution consumed is greater than the reduction in pore space volume, the pore pressure will decrease, whereas if the volume of aqueous solution suffers a smaller reduction than the pore volume, then the internal pressure will increase.
Because olivine alteration consumes water, a maximum degree of alteration for a closed system can be estimated for a given porosity. We estimated the maximum extent of olivine alteration in an isochemical closed system as a function of pore-volume fraction, assuming a fully water-saturated pore space and no lower limit for the water activity to stop the process. The initial conditions and the secondary phases were the same as in Table 1, except that no gas phase was defined. Figure 5 shows the maximum percentage of olivine that can be hydrated as a function of rock/water content given as initial pore volume fraction. The simulations suggest that, for a complete conversion of Fo90 to secondary minerals, ca. 36% water-saturated pore/fracture volume is required. Therefore, in a closed system at low water contents, because of either the presence of a gas phase reducing the water saturation or limited pore space, olivine can only be partly altered (Fig. 5).
We simulated the same system as in the previous paragraph and estimated the volume changes of the solid and aqueous phases and the maximum potential for olivine alteration as a function of initial pore-volume fraction. In this case, we also limited the alteration potential by including proper activity calculations for the water phase, using the PHREEQC code. The total volume of the system was constant, but pressure changes following the alteration were not estimated. Densities of olivine, the secondary mineral phases, and the aqueous phase used for the calculations are listed in Table 2. Figure 6 shows the volume changes of the aqueous phase and the pore space and the maximum percentage of Fo90 that can be hydrated. The simulations suggest that the volume change of the pore space is less than that of the aqueous phase and that the maximum alteration potential therefore is given by the amount of water present (Fig. 6a). In such a closed system, the hypothetical expansion of the water phase into the pore space would lead to lower pressures, which would promote fluid flow into the system or rheological collapse. For higher-temperature systems with lower water densities, the difference between the volume changes would be larger. Because the reduced water activities limited complete hydration in this simulation, the maximum alteration potential was lower (Fig. 6b) than for the previous simulation (Fig. 5).
We have demonstrated that SO42− may postpone generation of H2 and reduce the aqueous concentration of H2 by increasing the aqueous O2 activity. The effect was, however, suggested to be small, as SO42− was rapidly reduced at modest amounts of olivine dissolved (<0.1mol Fo90/kgw). was shown to be irreversibly reduced to N2 and should have little or no effect on the H2 generation. When it comes to the carbon species, aqueous Fischer-Tropsch–type reactions, such as Reactions 12 and 13, have served to explain hydrocarbon plumes from hydrothermal sources such as the Lost City and Rainbow hydrothermal fields where CH4 coexists with H2 (e.g., Holm and Charlou, 2001; Proskurowski et al., 2008). The reduction of CO2 or CO to CH4, however, is known to only occur at elevated temperatures and in the presence of catalysts, such as magnetite (e.g., Berndt et al., 1996). At high temperatures and in the presence of metal-bearing catalysts such as magnetite, awaruite, native Fe, and Ni-Fe compounds, H2 generation may be affected by the presence of CO2 in two ways: (1) CH4 formation from CO2 consumes H2 and drives the dissociation of water into H2 and O2 forward and (2) more ferrous iron needs to be oxidized as O2 is supplied from CO2. At low temperatures and without any bioactivity, the formation of CH4 from CO2 is likely to be limited (e.g., McCollom and Bach, 2009), and CO2 and CH4 will most likely coexist in metastable equilibria. If this is the case, the O2 activity given by Eq. 13 is no longer dependent on the activity ratio of CO2 and CH4, and the potential for H2 production may be unaffected by CO2. The effect of aqueous carbon, sulfur, and nitrogen on the potential for hydrogen generation is hence suggested to be minor. The other factors treated here are the dissolution rates and water availability to fresh surfaces in closed systems. It was shown that the fraction of pore or fracture volume and the amount of water determine the maximum amount of olivine altered in such systems. Because these constraints are only valid for closed systems, we show in the next section that the impact on the emergence of life is likely to be minimal, as one of the key assumptions for the emergence of life is serpentinization in open alkaline systems (e.g., Martin and Russell, 2007; Russell, 2007; Russell et al., 2010).
During the first millions of years after the accretion of Earth in the Hadean Eon, the surface of Earth was hot and hostile. After 200–400 million years, however, detrital zircons formed, which indicates liquid surface waters and surface temperatures below the boiling point of water (Mojzsis et al., 2001; Wilde et al., 2001; Watson and Harrison, 2005). Moreover, inversion tectonics (Cooper and Williams, 1989) or early plate tectonics (e.g., Harrison et al., 2005) were active and refreshed the rock surfaces in contact with liquid water. Earth was then heavily bombarded, which heated up the thin lithospheric shell at least locally. The Hadean ended after the Late Heavy Bombardment at 3.9–3.8Ga (Tera et al., 1974). Trace evidence of putative microbial life from shortly after the Late Heavy Bombardment indicates that life had already emerged (e.g., Mojzsis et al., 1996; Furnes et al., 2004).
The tectonics during Hadean Earth reworked crustal material, and mantle convection depressurized and partially melted the hot mantle at greater depths and higher pressures than modern analogues. This led to high-magnesium melts that formed extensive ultramafic intrusives and komatiitic lavas (Nisbet and Sleep, 2001). Komatiitic lavas resemble ultramafic minerals such as olivine, with high magnesium contents, low silica, and Fe2+ iron of ca. 10% (Smith et al., 1980). Indeed they are largely composed of olivine, often with spinifex texture. The global potential for H2 generation in the Hadean can thus be compared to the present potential by assuming that the ultramafic lavas and intrusives chemically corresponded to modern olivine-bearing ultramafic rocks. At present, ultramafic mantle rocks are mainly exposed to seawater along amagmatic sections of slow- to ultraslow-spreading mid-ocean ridges (Dick et al., 2003) and at faster-spreading ridges where deep conductive extensional faults and fractures occur (Baker et al., 1994). If we assume that plate tectonics operated during the Hadean Eon, the higher heat flow from the warmer mantle would likely have increased the partial melting of the mantle and led to extensive eruption of komatiitic lavas. It is therefore likely that, on a global scale, more ultramafic material was exposed to liquid water in the Hadean than at the present time.
The evolution of the atmospheric O2 level has strongly influenced the oxidation level of seawater. During the Hadean and Archean eons, the atmosphere was likely to have been moderately oxidized and dominated by CO2 and N2 with low levels of O2 (Shaw, 2008). Sometime between 2.45 and 2.22 billion years ago, the atmospheric O2 rose from the low levels to higher than 10−5 times the present atmospheric level (Bekker et al., 2004) and thereafter with a stepwise increase to the present level due to the advent of microbial photosynthetic activity. Because of the low O2 levels before 2.45–2.22 billion years ago, ocean SO42− was low with concentrations <200μM, with maximum values of 2–3 orders of magnitude lower than present-day seawater (Bekker et al., 2004). Thus, based on the previous sections on the effect of initial SO42− and O2 on the rate of H2 generation, we conclude that the Hadean and Archean atmospheres together with water chemistry promoted enhanced H2 generation compared to the present time.
Alkaline springs analog to the present-day Lost City hydrothermal field (Kelley et al., 2005) have been suggested as potential sites for the emergence of life (Martin and Russell, 2007; Russell, 2007; Russell et al., 2010). Such systems are driven by serpentinization where H2 and possibly light hydrocarbons form (Kelley et al., 2001, 2004, 2005; Früh-Green et al., 2004; Proskurowski et al., 2008). In these systems, hot alkaline fluids from deep underground meet cold acidic surface waters where stable steep thermodynamic gradients form. These thermodynamic gradients, combined with the availability of building blocks of life, are hypothesized to have provided favorable conditions for the emergence of chemolithoautotrophic microbial life (Russell et al., 1989, 2010; Kelley et al., 2002; Russell, 2007; Martin and Russell, 2007).
We further suggest that the availability of water and pore/fracture space are the most important factors that limit the dissolution of olivine and generation of H2. This has implications for the colonization and spreading of microbial life on early Earth and on other habitable planets and moons. From constraints on pore space and water volumes, we suggest that the bulk H2 generation should be in open systems such as permeable ultramafic sediments and fracture or fault zones, where fresh aqueous solutions can react for sufficient time to produce H2. From the olivine-bearing cumulates or lavas, H2 and H2-derived aqueous species can then migrate as diffuse flows through overlying permeable units to the seafloor (Fig. 7). Once formed, the microbial communities could slowly develop and colonize the permeable seafloor, and the colonies could be carried along by plate tectonics (Russell, 2007, and references therein). Alternatively, microbial communities protected in low-density biofilms could detach from the mounds and be carried further by ocean currents (Fig. 7). Finally, because of the larger heat flow, more abundant ultramafic rocks, larger thermodynamic gradients offered by the CO2-rich atmosphere, and more-reduced oceans, the young Earth probably offered more systems similar to the Lost City field where life could possibly have emerged.
Although recently questioned by Zahnle et al. (2010), it is generally accepted that CH4 is released in large quantities from regions on Mars (Formisano et al., 2004; Mumma et al., 2009). One indication of the source of the methane is given by recent numerical simulations that suggest CH4 is released in pulses a short time before observation and from a broad source region rather than from point sources (Mischna et al., 2011). One of the potential causes for CH4 formation on a regional scale is serpentinization that occurs beneath the surface of the planet (Oze and Sharma, 2005). The discovery of serpentine minerals has confirmed that hydration of olivine occurs, or has occurred, on Mars (Ehlmann et al., 2010). Indeed, Mars houses olivine minerals (Hoefen et al., 2003; Koeppen and Hamilton, 2008) and liquid water (Malin and Edgett, 2000) where possible hydrothermal processes (Marzo et al., 2010; Morris et al., 2010) may have provided the thermodynamic conditions for serpentinization to take place (Oze and Sharma, 2007; Quesnel et al., 2009). These geological settings are likely not unique to Earth and Mars and may be common to the other terrestrial bodies in the Universe.
Water, a requisite for H2 generation, is probably one of the most abundant molecules in the Universe (Kotwicki, 1991). It has been detected on a number of terrestrial planets and natural satellites, for example, Venus (Svedhem et al., 2007), Mercury (Zurbuchen et al., 2008), Titan (Griffith et al., 2003), and Earth's moon (Feldman et al., 1998), although mainly in the form of ice. Evidence of the existence of liquid water, which is essential for serpentinization and life on Earth, has been reported on Mars, Europa (Carr et al., 1998), Enceladus (Waite et al., 2009), and Titan (O'Brien et al., 2005). Furthermore, liquid water is proposed to have existed on many extraterrestrial bodies at certain periods of their evolutionary history (Kasting, 1988; Baker et al., 2005).
Olivine and its high-pressure equivalents are believed to be the major mineral components of the mantles of all Earth-like planets and moons (Sohl et al., 2002). The main constituent of pallasite, a differentiated meteorite whose composition reflects the interiors of parent bodies, is olivine (Buseck, 1977). Olivine has been identified, for example, on Mars (Papike et al., 2009), Mercury (Sprague et al., 2009) and Earth's moon (Lucey, 2004). Driven by the heat transfer from the mantle to the surface during the cooling of the Earth-like planets and satellites, hydrothermal processes are likely to operate when water is present (Vance et al., 2007). Analysis of carbonate minerals in a meteorite from ancient Mars has indicated that this process happened on early Mars (Treiman et al., 2002). Based on the differential structure that consists of a hot rocky core and a water shell, and the observation of its active region on the south pole, Glein et al. (2008) asserted that such a cooling-induced hydrothermal process also occurred on early Enceladus. In addition, impact-induced hydrothermal activity is an important process in our Solar System (Koeberl and Reimold, 2004; Abramov and Kring, 2005).
Based on the availability of liquid water, olivine, and local hydrothermal activity, we conclude that H2 generation and possible conditions suitable for the emergence of life may be, or may have been, present on Mars and other Earth-like planets and moons. H2 is the strongest electron donor possible to provide energy for microorganisms on Earth. Here, we have provided model calculations to demonstrate that, with suitable rocks and conditions, subsurface H2 production is possible elsewhere. Furthermore, the discovery of hydrogen peroxide (H2O2) on the surface and in the atmosphere of Mars (Clancy et al., 2004; Encrenaz et al., 2004) suggests that the most habitable environment on Mars would be deep in the subsurface where the activity of H2O2 is zero. There, the potential for H2 generation through serpentinization of olivine in low-temperature systems will be limited by the same factors as reviewed in this paper.
As H2 is a key ingredient in the emergence of life and a basal part of the deep biosphere, knowledge of the potential for H2 generation from water-rock interactions has received considerable attention. Here, we have provided a review of factors that may reduce the potential for the generation of H2 with focus on low-temperature systems. We have shown that sulfate may affect the potential for H2 generation as it potentially constrains the aqueous O2 at levels higher than required to achieve high H2 concentrations. Redox-sensitive compounds of carbon and nitrogen, on the other hand, are unlikely to have significant effect at low temperatures. We further suggest that the availability of water and pore/fracture space are the most important factors that limit the dissolution of olivine and generation of H2. Finally, our study implies that, because of large heat flows, abundant olivine-bearing rocks, large thermodynamic gradients, and reduced atmospheres, the young Earth and Mars probably offered abundant systems where microorganisms could possibly have emerged.
We highly appreciate the reviewers and editors for improving the grammar and the clarity and vision of the manuscript. We are thankful to Eva Bjørseth at Department of Earth Science (University of Bergen) for improving the quality of our summary figure, and finally we would like to express our gratitude for fruitful discussions with Dr. Wolf Geppert and the other astrobiologists associated with the Swedish astrobiology graduate school.
kgw, kilogram of water.
No competing financial interest exists.