Reactions between komatiite and CO₂-rich seawater at 250 and 350 °C, 500 bars: implications for hydrogen generation in the Hadean seafloor hydrothermal system

Abstract

To understand the chemical nature of hydrothermal fluids in the komatiite-hosted seafloor hydrothermal system in the Hadean, we conducted two hydrothermal serpentinization experiments involving synthetic komatiite and a CO₂-rich acidic NaCl fluid at 250 and 350 °C, 500 bars. During the experiments, the komatiites were strongly carbonated to yield iron-rich dolomite (3–9 wt.% FeO) at 250 °C and calcite (<0.8 wt.% FeO) at 350 °C, respectively. The carbonation of komatiites suppressed H₂ generation in the fluids. The steady-state H₂ concentrations in the fluid were approximately 0.024 and 2.9 mmol/kg at 250 and 350 °C, respectively. This correlation between the Fe content in carbonate mineral and the H₂ concentration in the fluid suggests that the incorporation of ferrous iron into the carbonate mineral probably limited magnetite formation and consequent generation of hydrogen during the serpentinization of komatiites. The H₂ concentration of the fluid at 350 °C corresponds to that of modern H₂-rich seafloor hydrothermal systems, such as the Kairei hydrothermal field, where hydrogenotrophic methanogens dominate in the prosperous microbial ecosystem. Accordingly, the high-temperature serpentinization of komatiite would provide the H₂-rich hydrothermal environments that were necessary for the emergence and early evolution of life in the Hadean ocean. In contrast, H₂-rich fluids may not have been generated by serpentinization at temperatures below 250 °C because carbonate minerals become more stable with decreasing temperature in the komatiite-H₂O-CO₂ system.

Introduction

Deep-sea hydrothermal environments have been considered as a favorable environment for the emergence and early evolution of life on Earth (e.g., Yanagawa and Kojima 1985; Russell and Hall 1997). In particular, H₂-rich hydrothermal fluids generated through the serpentinization of ultramafic rocks would have driven prebiotic chemical evolution and the development of biotic energy metabolisms (Takai et al. 2006; Amend and McCollom 2009; Russell et al. 2014; Nakamura and Takai 2014; Shibuya et al. 2016). Molecular hydrogen (H₂) generation during serpentinization is caused by the reduction of water in conjunction with the oxidation of ferrous iron in silicates. This process is written as the following simplified reaction (e.g., Allen and Seyfried 2003; Seyfried et al. 2007; McCollom and Bach 2009):

$$ 2{\left(\mathrm{F}\mathrm{e}\mathrm{O}\right)}_{\mathrm{rock}} + {\mathrm{H}}_2\mathrm{O}\to {\left({\mathrm{Fe}}_2{\mathrm{O}}_3\right)}_{\mathrm{rock}} + {\mathrm{H}}_2, $$

where the ferric iron-bearing solid phase generally precipitates as magnetite (Fe₃O₄). Therefore, H₂ generation is strongly related to magnetite formation during serpentinization (McCollom and Bach 2009).

Among the various prospects for the seafloor hydrothermal systems in the early Earth (e.g., Kump and Seyfried 2005; Shibuya et al. 2010), two types of serpentinization are hypothesized to be dominant in the Hadean ocean: a komatiite-hosted high-temperature type at oceanic islands/plateaus (Takai et al. 2006; Yoshizaki et al. 2009; Shibuya et al. 2015) and a low-temperature alkaline type hosted by komatiites or peridotites at off-ridge ocean floors (e.g., Russell et al. 2010, 2014; Shibuya et al. 2016). Based on the geological records, Archean oceanic crust was likely much thicker than modern equivalents owing to the higher potential mantle temperature at that time (Ohta et al. 1996; Komiya et al. 2002; Komiya 2004; Moores 2002; Shibuya et al. 2012). This thick lid of oceanic crust probably limited the exposure of mantle peridotites on the seafloor, thus suggesting that komatiite volcanism would have been much more abundant than the exposed mantle peridotites that are frequently observed near modern slow-spreading ridges without a sufficient magmatic supply (Takai et al. 2006). Therefore, it has been pointed out that H₂-rich seafloor hydrothermal environments would have been predominantly driven by komatiite volcanism in the Hadean ocean (Takai et al. 2006). On the other hand, the low-temperature alkaline type could provide distinctive chemical environments such as large pH gradients in the seawater-hydrothermal fluid mixing zones, which may also have been advantageous for the possible development of proton-motive energy metabolisms (Russell et al. 2014; Shibuya et al. 2016).

Previously, hydrothermal alteration experiments have been conducted to understand the potential of H₂ generation during the serpentinization of komatiites under CO₂-free conditions (Yoshizaki et al. 2009; Shibuya et al. 2015; Suzuki et al. 2015). However, as suggested by theoretical calculations (Walker 1985; Kasting 1993; Sleep and Zahnle 2001) and geological records (Lowe and Tice 2004; Ohmoto et al. 2004; Shibuya et al. 2007, 2012, 2013a), the atmospheric CO₂ levels in the early Earth were likely much higher than the modern level. Furthermore, it has been experimentally demonstrated that carbonate formation during the serpentinization of olivine under CO₂-rich conditions suppresses H₂ generation in fluids (Jones et al. 2010; Klein and McCollom 2013; Neubeck et al. 2014). Recently, some experiments have been conducted in komatiite-CO₂-H₂O systems (Lazar et al. 2012; Hao and Li 2015), but H₂ generation in fluids was not the objective in these studies. Therefore, the potential for hydrogen generation through the serpentinization of komatiites in the Hadean has not yet been experimentally evaluated under CO₂-rich conditions.

In this study, we conducted two experiments that simulate the reactions between komatiite and CO₂-rich seawater at 250 and 350 °C, 500 bars, using a batch-type (closed system) hydrothermal reactor (Yoshizaki et al. 2009). The experiments revealed the chemical composition of high-temperature hydrothermal fluids and CO₂ absorption ability of komatiite through serpentinization under CO₂-rich conditions. The results demonstrate the significance of komatiite-hosted hydrothermal systems for seawater chemistry and provide insight into the H₂-rich hydrothermal environments in the CO₂-rich Hadean ocean.

Methods/Experimental

Synthesis of komatiite

The komatiite used in the experiments was synthesized from a mixture of 12 reagents (SiO₂, TiO₂, Al₂O₃, Fe₂O₃, MnO, MgO, CaCO₃, Na₂CO₃, K₂CO₃, P₂O₅, NiO, and Cr₂O₃). The chemical composition of the mixed reagents was adjusted to Al-depleted (Barberton-type) komatiite (Smith et al. 1980; Wei et al. 1990; Yoshizaki et al. 2009) because the volcanism of Al-depleted komatiite was likely more predominant than Al-undepleted komatiite in the Hadean (Shibuya et al. 2015). The mixed powder was placed in a Pt-Rh crucible at 1000 °C for 1 h in an electronic furnace to decarbonate the reagents. The mixture was melted at 1600 °C for 0.5 h at the oxygen fugacity of quartz-fayalite-magnetite (QFM) buffer regulated by a H₂-CO₂ gas mixture (Canil 1997). To create a spinifex texture of olivine, the temperature was lowered to 1450 °C over 1.5 h. Next, the crucible was quenched to room temperature, thus yielding a fresh spinifex-textured komatiite (Fig. 1). The composition and texture of the synthetic komatiite and its mineral/glass phases were analyzed using X-ray diffraction (XRD), X-ray fluorescence (XRF), and an electron probe microanalyzer (EPMA) (Table 1). In the interstitial glass, dendritic olivine crystals configurate a spinifex texture. Broadly depending on size, these crystals have Mg# values ranging from 89 to 94. Small amounts of chromian spinels are disseminated in the vicinity of the olivine crystals. Owing to such a mineral assemblage, almost all Ca in the komatiite was partitioned into the interstitial glass (Fig. 1 and Table 1). The synthesized komatiite was crushed in a tungsten mortar and sieved to obtain a <100 μm powder. To remove any contamination of organic materials during sample preparation, the powdered komatiite was washed with acetone and distilled water several times and dried for 12 h in an oven at 50 °C.

Fig. 1

a Photomicrograph taken in cross-polarized light and b a backscattered electron (BSE) image of the synthetic komatiite. Ol olivine, Spl spinel, Gl glass

Table 1 Compositions of the synthetic komatiite, mineral/glass phases therein, starting materials used in a previous experiment (wt.%)

Experimental system

The Inconel alloy autoclave used for the hydrothermal alteration experiment is based on the study by Seyfried et al. (1979). It resists corrosion and withstands high-temperature and high-pressure conditions (up to 600 °C and 600 bars; Fig. 2). The reaction cell is made of a gold bag with a titanium head because of the resistance of these materials to high-temperature fluids. Furthermore, gold has sufficient flexibility to allow the fluid inside the reaction cell to be pressurized by the surrounding water. In addition, to avoid possible H₂ generation via the reaction of metallic titanium with water, the surface of the titanium was completely oxidized before the experiments. To eliminate organic contamination, all materials in contact with the reacting fluid were baked in a muffle furnace at 450 °C for 5 h.

Fig. 2

Schematic illustration of the Inconel alloy autoclave used in this study (modified after Yoshizaki et al. 2009). The synthesized komatiite and hypothetical Hadean seawater were sealed in the gold bag

The initial solution used in the experiments was controlled at pH 4.9 and 400 mmol/kg of ΣCO₂ (=CO_{2 (aq)} + HCO₃ ⁻ + CO₃ ²⁻) by the addition of NaHCO₃, NaCl, and HCl to ultrapure water, according to the previous studies suggesting that Hadean seawater was acidic due to the high partial pressure of CO₂ in the atmosphere (e.g., Kasting 1993; Macleod et al. 1994). To avoid the degassing of CO₂ by a reaction between NaHCO₃ and HCl prior to the sealing of the reaction cell, these materials were separately placed in the gold tube and then mixed after sealing (Shibuya et al. 2013b). The Cl concentration of the solution was regulated to be approximately 1000 mmol/kg because salinity of seawater in the early Earth was potentially 1.5–2 times higher than the modern value (approximately 550 mmol/kg) because of the absence of continents and the associated salt deposits/saline water (Knauth 2005). The komatiite powder was reacted with this hypothetical seawater in the reaction cell at 250 and 350 °C, 500 bars for 1530–2760 h (hereafter abbreviated as Exp-250 and Exp-350, respectively). At the beginning of the experiments, the water/rock mass ratios were adjusted to approximately 5 because water/rock ratios are commonly below 5 in the high-temperature regions of natural subseafloor hydrothermal systems (Wetzel and Shock 2000). The water/rock mass ratio decreased to ~3 at the end of the experiments due to the multiple fluid samplings during the experiments.

Sampling and analytical methods

During the experiment, approximately 3–4 g of fluid samples was collected several times through a gold-lined sampling tube. For CO₂ and H₂ analyses, 0.5 mL of fluid sample was directly introduced to each Ar-purged, sealed vial at room temperature. Especially, for the CO₂ analysis, the sampled fluid was acidified (pH <2) by adding HCl to ensure the complete extraction of the dissolved bicarbonate and carbonate ions. Quantitative analysis for gas species was conducted by gas chromatography. The overall detection precision for the analyses of ΣCO₂ and H₂ concentrations in the fluid were both better than 5%. For the analysis of other dissolved species, 0.2 mL of the fluid was collected in two vials, to which either HNO₃ or NaOH was immediately added to avoid the precipitation of transition metals (Fe and Mn) and silicic acid species, respectively. The fluid samples were analyzed using inductively coupled plasma optical emission spectrometry and ion chromatography. The analytical precision (2σ) was approximately 2% for Cl and Na, and 5% for the other elements.

The pH of the fluid samples was determined using a pH meter at 25 °C under atmospheric conditions. The significant digits and precision of the pH measurement are 2 digits and ±0.1 units. This measurement was performed 1 h after sampling to allow the stabilization of the pH against CO₂ degassing. The pH_25 °C of the fluid cannot be measured directly because it rapidly changes due to CO₂ degassing after the sampling of fluid from the reactor. The amount of degassed CO₂ was determined from the ΣCO₂ concentrations of the fluid sample directly introduced into the sealed vial and of the fluid degassed under atmospheric conditions, then the original pH_25 °C was calculated based on the amount of degassed CO₂. We calculated pH_in-situ of the high-temperature fluid in the reaction cell with the Geochemist’s Workbench computer code (Bethke 2008) based on the pH at room temperature (pH_25 °C) and concentrations of dissolved elements/species and gases. In the pH_in-situ calculations, charge balance was constrained by pH_25 °C, while Na was used to compensate for imbalanced charges derived from analytical errors. The required thermodynamic database was generated by the SUPCRT92 computer program (Johnson et al. 1992), using thermodynamic data of minerals, aqueous species, and complexes reported in Shock and Helgeson (1988), Shock and Koretsky (1995), Shock et al. (1989, 1997), Sverjensky et al. (1997), and McCollom and Shock (1997). The B-dot activity model was used in the calculations (Helgeson 1969; Helgeson and Kirkham 1974). The temperature-dependent activity coefficient for aqueous CO₂ was derived from the empirical relationship established by Drummond (1981), and the temperature-dependent activity of water in NaCl solution was derived from the formulation of Bethke (2008). Cleverley and Bastrakov (2005) provided useful temperature-dependent polynomial functions for both of these last-mentioned parameters.

After the experiments, alteration products were retrieved from the reaction cell and dried at 80 °C for 12 h. The solid alteration product of each experiment was analyzed with XRD and EPMA to determine the assemblage and composition of the alteration minerals. Powdery alteration products from the experiments at both temperatures were embedded in Epofix resin, and thin sections were made of these. The analytical conditions of the EPMA were 15 kV of accelerating voltage, 10 nA of specimen current, and 60–80 s of counting time. The precision of the elemental concentration was less than 10%, which was based on duplicate measurements. The detection limit for the elemental concentration was 0.01 wt.%. Moreover, the starting solid material (unaltered komatiite) was analyzed with XRD, XRF, and EPMA.

Results and Discussion

Fluid chemistry

The Cl concentration in the fluid was kept relatively constant throughout both experiments because Cl-bearing minerals are generally rare in hydrothermally altered rocks (Fig. 3 and Table 2). In contrast, the concentrations of all other components (Na, K, Mg, Ca, Si, Fe, Mn, and ΣCO₂) changed immediately after the beginning of the experiments and reached a near steady state within 2760 h at 250 °C and 1530 h at 350 °C. The Mg concentration in Exp-250 (36–40 mmol/kg) was 30–40 times higher than that in Exp-350 (<1.2 mmol/kg). The Fe concentration in Exp-250 (0.49–0.97 mmol/kg) was approximately half than that in Exp-350 (1.53–1.80 mmol/kg). No large difference between Exp-250 and Exp-350 was identified in the concentrations of other dissolved elements. Na (820–1036 mmol/kg across both experiments) had the second highest concentration next to Cl. The concentration of Ca increased and converged at 39–49 mmol/kg in Exp-250 and 37–42 mmol/kg in Exp-350. The concentrations of Si and K were 1.6–6.1 and 0.78–1.24 mmol/kg, respectively, across both experiments. The concentration of Mn was the lowest among all analyzed elements (0.15–0.73 mmol/kg across both experiments).

Fig. 3

Concentrations of the dissolved species in an aqueous fluid coexisting with komatiite powder and its alteration products as a function of reaction time in the experiments at a 250 and b 350 °C, 500 bars

Table 2 Composition of the sampled fluids in the experiments (mmol/kg)

The ΣCO₂ concentration decreased from ca. 400 mmol/kg at the beginning of the experiments to approximately 33 mmol/kg in Exp-250 and approximately 171 mmol/kg in Exp-350 (Fig. 4a and Table 2). This trend of the steady-state ΣCO₂ concentration in the higher-temperature experiment being higher than that in the lower-temperature experiment is consistent with the results of experiments simulating high-temperature reactions between basalt and CO₂-rich seawater (Shibuya et al. 2013b). On the other hand, the pH_in-situ values were higher than the pH_25 °C values at the beginning of the experiments due to the temperature dependence of the ion product of water and speciation of CO₂ species (e.g., Shibuya et al. 2010). As the reaction proceeded, the pH_in-situ values finally decreased to 4.8 in Exp-250 and 5.7 in Exp-350, which are broadly consistent with those of high-temperature hydrothermal fluids in modern oceans (pH_in-situ = approximately 5) (Seyfried et al. 1991).

Fig. 4

Changes in a dissolved CO₂ and b H₂ concentrations in the sampled fluids as a function of reaction time

In both experiments, the H₂ concentration in the fluid suddenly increased and temporarily reached a high value just after the start of the experiments, then slightly decreased in each experiment (Fig. 4b). Subsequently, the H₂ concentration remained almost constant in Exp-350, whereas it gradually increased with time in Exp-250. Finally, it converged to reach a steady state at up to 0.024 mmol/kg in Exp-250 and 2.9 mmol/kg in Exp-350 (Table 2). This trend indicates that the H₂ concentration was unstable due to extensive water/rock reactions just after the start of the experiments, and then gradually reached a steady state. The steady-state H₂ concentration in Exp-350 was approximately 100 times higher than that in Exp-250, thus indicating that differences in temperature significantly affect hydrogen generation during water-rock reactions.

Alteration products

The XRD analysis revealed that the synthetic komatiites were extensively altered throughout the experiments (Fig. 5). The results show that the alteration product in Exp-250 contains dolomite, serpentine, and a smectite/chlorite mixture (Fig. 5b), whereas calcite, serpentine, and a smectite/chlorite mixture were present as major alteration minerals in Exp-350 (Fig. 5c). The existence of serpentine and carbonate minerals in the run products indicated komatiite was altered by both serpentinization and carbonation. The precipitated carbonate mineral differed for the two temperatures.

Fig. 5

Representative XRD patterns (CuKα) of a the unaltered synthetic komatiite and the alteration products of b Exp-250 and c Exp-350. Dol dolomite, Cal calcite, Sme smectite, Chl chlorite, Ol olivine, and Serp serpentine

Secondary electron imaging (SEI) and EPMA spot analysis provided more detailed information on mineral assemblage and composition (Fig. 6 and Table 3). In Exp-250, EPMA analyses of the products revealed that dolomite crystals contained a relatively high content of iron (3–9 wt.% FeO_total), whereas the serpentine and smectite/chlorite mixture had FeO_total contents of 13–15.5 wt.% (Fig. 6a–d; Table 3). Siderite (including approximately 40 wt.% FeO_total) and olivine were present as minor minerals in the alteration products in Exp-250 (Fig. 6a, c; Table 3). In Exp-350, the EPMA analyses of the alteration products revealed that calcite crystals had a relatively low FeO_total content (0.1–0.8 wt.%) and that the serpentine and smectite/chlorite mixture had FeO_total contents of 14–17 wt.% (Fig. 6e; Table 3). Olivine and magnetite were also identified as minor minerals; however, according to the SEI observations in many places over the thin sections, their abundances were likely greater in Exp-350 than in Exp-250. Comparing the compositions of serpentine and the smectite/chlorite mixture between Exp-250 and Exp-350, no large difference in FeO_total content was identified. On the other hand, on average, the FeO_total content in Fe-bearing dolomite in Exp-250 was ten times higher than that in calcite in Exp-350 (Table 3).

Fig. 6

Representative SEI images of alteration products (a–d) in the experiment at 250 °C and (e, f) in the experiment at 350 °C. Sd siderite, Dol dolomite, Cal calcite, Sme smectite, Chl chlorite, Ol olivine, Serp serpentine, and Mag magnetite

Table 3 Composition of run products (wt.%)

Dissolved CO₂ concentration and pH

The ΣCO₂ concentrations in fluids decreased from approximately 400 to 33 mmol/kg in Exp-250 and 171 mmol/kg in Exp-350 (Fig. 4a and Table 2). These differing steady-state ΣCO₂ concentrations indicate that carbonate minerals are destabilized with increasing temperature in the komatiite-CO₂-H₂O system. This trend is consistent with previous basalt experiments under CO₂-rich conditions at 250 and 350 °C, where the final ΣCO₂ concentrations were 1 mmol/kg at 250 °C and 108 mmol/kg at 350 °C (Shibuya et al. 2013b). Comparing results for the same temperature, the ΣCO₂ concentration in our komatiite experiments is higher than that in the basalt experiments of Shibuya et al. (2013b), thus indicating that komatiite has a lesser ability than basalt to absorb CO₂ through water-rock reactions. This would be derived from the lower CaO content of komatiite (compared with basalt) because CaO is the main component in the alteration carbonate (mainly calcite or dolomite) in the basalt and komatiite experiments (Shibuya et al. 2013b and the present study).

The final pH_in-situ values of fluids in the present experiments (pH_in-situ = 4.8 at 250 °C and 5.7 at 350 °C) are clearly lower than those of the previous basalt-CO₂-NaCl fluid experiments (pH_in-situ = 6.6 at 250 °C and 7.2 at 350 °C) (Shibuya et al. 2013b). This discrepancy is probably derived from the lower pH_25 °C of the initial solution (pH_25 °C = 4.9) in our experiments than the value of 6.5 in the basalt experiments and/or compositional differences between komatiite and basalt. In general, under CO₂-rich and near-neutral pH conditions at room temperature, substantial HCO₃ ⁻ is dissolved in the solution. With increasing temperature, the dissolved HCO₃ ⁻ is converted to CO_2(aq) and OH⁻, which elevates the pH_in-situ value of the hydrothermal fluid (Shibuya et al. 2010). Although reactions with rocks generally cause fluid pH_in-situ to converge near 5.5 (neutral pH at 250–350 °C), previous experiments have revealed that the alkalinization effect due to the increase in temperature under CO₂-rich conditions potentially elevates the pH_in-situ value beyond a neutral pH value at high temperature, even if the fluid reacts with rocks (Shibuya et al. 2013b). In this study, however, the assumed pH_25 °C of the initial solution was 4.9, which was likely too low to form substantial HCO₃ ⁻ in the initial solution at room temperature. This implies that if seawater pH was lower than approximately 5 (CO_2aq was the main species), metal-rich black smoker-type hydrothermal fluids would have been generated in the Hadean komatiite hydrothermal systems in contrast to the metal-poor, alkaline hydrothermal fluids in the Archean basalt-hosted system. Although the difference in whole rock composition between komatiite and basalt should be experimentally evaluated, seawater pH was likely an important factor controlling the composition of hydrothermal fluids in the early Earth.

Serpentinization of komatiites

In this section, we compare our experimental results with conditions in natural environments and thermodynamic calculations in order to characterize the reaction occurring in the reaction cell. The EPMA analysis of alteration products revealed the presence of a small quantity of olivine in both experiments, but the amount of olivine in Exp-350 was greater than that in Exp-250. Thermodynamic calculations for serpentinization of olivine and harzburgite (80 wt.% of olivine, 15 wt.% of orthopyroxene, and 5 wt.% of clinopyroxene) indicate that olivine becomes stable and is one of the major minerals at temperatures above approximately 315–350 °C in the peridotite- or olivine-H₂O system (e.g., McCollom and Bach 2009; Klein et al. 2013). In analogy with the calculation, the experiments in this study suggest that olivine stabilizes with increasing temperature, even during komatiite serpentinization at 250–350 °C. Nevertheless, olivine was observed as a minor phase even in Exp-350. This is probably due to the composition of komatiite that contains a certain level of Al₂O₃ would stabilize serpentine and/or chlorite/smectite because smectite becomes stable as bulk Al₂O₃ content increases in ultramafic rocks such as komatiite (Shibuya et al. 2015). However, it was difficult to determine whether the igneous olivine still remains in the absence of any reaction with a fluid or if the alteration olivine was newly created through serpentinization.

Brucite commonly occurs in serpentinized rocks in modern oceanic plates (e.g., Bach et al. 2006; Klein et al. 2014) but was not found in the alteration products in the present experiments. The absence of brucite in Exp-350 is consistent with the thermodynamic calculations for the serpentinization of olivine and harzburgite under CO₂-poor conditions, since brucite is destabilized at temperatures above ~350 °C (McCollom and Bach 2009; Klein et al. 2013). On the other hand, the instability of brucite in Exp-250 is likely due to the CO₂-rich condition in the experiment because previous experiments revealed that Mg-bearing carbonate mineral is more stable than brucite under CO₂-rich conditions (Zhao et al. 2010; Hövelmann et al. 2012; Klein and McCollom 2013). Alternatively, high dissolved Si content in the fluid (3.7–4.1 mmol/kg) possibly prohibited brucite formation, which is supported by thermodynamic calculations (e.g., Klein and McCollom 2013).

Carbonation of komatiites

The steady-state ΣCO₂ concentrations in both experiments indicate that the amount of carbonate mineral was greater in Exp-250 than in Exp-350 (Fig. 4a and Table 2). The formation of dolomite in Exp-250 is characteristically different from the basalt-CO₂ experiments that yielded only calcite as carbonate in previous studies at 250 and 350 °C (Shibuya et al. 2013b). Comparing the experimental conditions of this study to those in the previous study, compositional differences between komatiite and basalt and/or the difference in initial pH (4.9 in this study, and 6.5 in Shibuya et al. (2013b)) likely caused the difference in precipitated carbonate species in the alteration products at 250 °C. Furthermore, it was revealed that the precipitated carbonate species strongly affected the Mg concentration in the hydrothermal fluid, i.e., Mg concentration in Exp-250 (carbonate as dolomite) is 30–40 times higher than that in Exp-350 (carbonate as calcite), while fluid Ca concentrations in both experiments are similar.

Previously, carbonate species were reported to correlate with the concentration ratio between Mg and Ca (mMg²⁺/(mMg²⁺ + mCa²⁺)) and the temperature of the fluid (Fig. 7) (Rosenberg and Holland 1964; Rosenberg et al. 1967; Tribble et al. 1995). In Exp-350, carbonate was precipitated as calcite, whereas the mMg²⁺/(mMg²⁺ + mCa²⁺) value of the fluid was approximately 0.03; thus, it clearly falls within the stability field of calcite. On the other hand, the mMg²⁺/(mMg²⁺ + mCa²⁺) value of the fluid in Exp-250, where dolomite precipitated, was approximately 0.43, which is near the dolomite/magnesite stability boundary. Therefore, the extremely high fluid Mg concentration in Exp-250 is consistent with the precipitation of dolomite in the alteration product. However, it is difficult to discuss the anteroposterior relation between dolomite formation and enrichment of Mg in fluids because the decrease in ΣCO₂ concentration and the Mg enrichment in the fluid were already confirmed at the first sampling. It is widely accepted that seafloor hydrothermal systems play a significant role as Mg sinks in the modern ocean because basaltic oceanic crusts incorporate seawater Mg into alteration minerals such as smectite and chlorite through hydrothermal alteration at near mid-ocean ridges (e.g., Alt 1995). In addition, Charlou et al. (2002) reported that ultramafic rock-hosted high-temperature hydrothermal systems provide fluids with little Mg into seawater and generally act as Mg sinks in the ocean. On the other hand, weathering of ultramafic rocks exposed at seafloor below 150 °C and hydrothermal circulation at high water/rock ratios provide Mg into seawater (Snow and Dick 1995). Our results hypothesize that dolomite formation during the serpentinization of komatiite would be a source of Mg for the Hadean ocean, even when the temperature range of hydrothermal reaction zones was near 250 °C. Further experiments will be necessary to justify this hypothesis.

Fig. 7

The mMg²⁺/(mMg²⁺ + mCa²⁺) values of hydrothermal fluids in Exp-250 and Exp-350 plotted on the calcite-dolomite-magnesite stability fields estimated from experiments at 275–420 °C and their extrapolation to 0 °C (modified after Tribble et al. 1995). The dotted and solid curves are for the solutions with 2 and 1 M of Ca-Mg chloride, respectively. The blue circle and pink diamond indicate the values in Exp-250 and Exp-350, respectively

The dolomite crystals formed in this study contain up to 9 wt.% FeO_total in Exp-250 because Fe substitutes Mg in dolomite during its precipitation owing to the similarity in ion radius between Fe and Mg (Mg²⁺: 0.72 Å; Fe²⁺: 0.61 Å) (Jia 1991). A similar Fe-bearing carbonate mineral was also reported in a hydrothermal alteration experiment for komatiite conducted under CO₂-rich conditions (water 50.1 mg + CO₂ 7.3 mg: about 3300 mmol/kg at the starting condition) at 300 °C and 500 bars (Hao and Li 2015). Furthermore, Klein and McCollom (2013) reported the precipitation of Fe-bearing magnesite (FeO = 6.78 wt.%) through the serpentinization of olivine under CO₂-rich conditions at 230 °C. Although differences in the composition of the starting materials between the previous studies and our experiments would have caused the difference in carbonate species, it seems likely that carbonate minerals formed during serpentinization under CO₂-rich conditions contain substantial FeO, probably at temperatures below approximately 300–350 °C.

Effect of seawater CO₂ on H₂ concentration in fluid

The maximum steady-state fluid H₂ concentrations were 0.024 mmol/kg in Exp-250 and 2.9 mmol/kg in Exp-350 (Fig. 4b), and the results are consistent with the analysis of solid alteration products. Magnetite, whose formation is related to H₂ generation (McCollom and Bach 2009), was observed in Exp-350 but not in Exp-250. Nevertheless, the H₂ concentrations in both experiments were clearly lower than the steady-state H₂ concentration (~20 mmol/kg) in fluid generated by the reactions between komatiite and a CO₂-free NaCl fluid at 300 °C and 500 bars (Shibuya et al. 2015). Although direct comparison is difficult owing to the difference in temperature conditions, two possible factors explain the differences between the previous and present experiments in terms of the steady-state H₂ concentrations from the hydrothermal reaction of komatiite.

The first factor is the temperature dependence of the stability of magnetite in ultramafic systems. At temperatures below 315–350 °C, the amount of magnetite formation increases with increasing temperature, but it drastically decreases at temperatures above 315–350 °C because olivine becomes stable (McCollom and Bach 2009; Klein et al. 2013). Moreover, Klein et al. (2013) reported that within the temperature range of 25–400 °C, the amount of hydrogen generation caused by the reaction with olivine (Fo 90) reaches a maximum at 322 °C. Thus, the temperature dependence of the stability of olivine and magnetite would be a primary cause of the difference in fluid H₂ concentrations between the previous CO₂-free komatiite experiment at 300 °C and our experiments.

The other factor is the presence of CO₂ in the experimental system. Based on thermodynamic calculations simulating the relative amount of magnetite formation at 50–400 °C under 500 bars, Klein et al. (2013) expected that the amount of hydrogen generation caused by the reaction with olivine under CO₂-free conditions is greater at 250 °C than at 350 °C. In our experiments, however, the steady-state fluid H₂ concentration was higher in Exp-350 than in Exp-250. In previous experiments that simulated the reactions between olivine and a CO₂-rich fluid at 230 °C and 35 MPa, ferrous iron-bearing talc and magnesite were generated, while magnetite formation and H₂ generation were limited (Klein and McCollom 2013). Under CO₂-free conditions, the iron is only partitioned into magnetite and ferrous iron-bearing silicate in alteration products,

$$ \mathrm{a}{\left(\mathrm{F}\mathrm{e}\mathrm{O}\right)}_{\mathrm{rock}}+{\mathrm{bH}}_2\mathrm{O}\to \mathrm{d}{\left({\mathrm{Fe}}_2{\mathrm{O}}_3\right)}_{\mathrm{rock}}+d{\mathrm{H}}_2+\left(\mathrm{a}\hbox{-} 2d\right){\left(\mathrm{F}\mathrm{e}\mathrm{O}\right)}_{\mathrm{rock}.}. $$

In contrast, the interaction between CO₂-rich fluid and rock leads to carbonate mineral precipitation and ferrous iron incorporation therein. The reaction occurring under a CO₂-rich condition is the following:

$$ \mathrm{a}{\left(\mathrm{F}\mathrm{e}\mathrm{O}\right)}_{\mathrm{rock}}+{\mathrm{bH}}_2\mathrm{O}+{\mathrm{cCO}}_2\to \mathrm{d}{\left({\mathrm{Fe}}_2{\mathrm{O}}_3\right)}_{\mathrm{rock}}+{\mathrm{dH}}_2+\mathrm{e}{\left(\mathrm{F}\mathrm{e}\mathrm{O}\right)}_{\mathrm{rock}}+\left(\mathrm{a}\hbox{-} 2d\hbox{-} \mathrm{e}\right){\left({\mathrm{Fe}\mathrm{CO}}_3\right)}_{\mathrm{rock}}+\left(\mathrm{c}\hbox{-} \mathrm{a}+2d+\mathrm{e}\right){\mathrm{CO}}_2. $$

In both equations, the value of “d” is the amount of H₂ generation and depends on temperature. Therefore, it was suggested that the lack of H₂ generation is due to the direct incorporation of ferrous iron into carbonate minerals without its oxidation during the hydrothermal alteration. In our experiments, although a large difference in the composition of serpentine and smectite/chlorite mixture was not identified between Exp-250 and Exp-350, the Fe content of dolomite in Exp-250 (3–9 wt.% FeO_total) was clearly higher than that of calcite in Exp-350 (Table 3). The amount of generated magnetite in Exp-350 was greater than that in Exp-250. These data suggest that the amount of ferrous iron incorporated into the carbonate mineral strongly affected the extent of magnetite formation and resulting hydrogen generation in the fluid. Although such a trend in the effect of CO₂ on H₂ generation has been reported in some previous experiments at temperatures below 230 °C (Jones et al. 2010; Klein and McCollom 2013; Neubeck et al. 2014), our experiments show that the suppression of H₂ generation by the presence of CO₂ in the system occurs even at higher temperatures (250 °C) and in the CO₂-H₂O-komatiite system. In addition, a small amount of ferrous iron incorporated into calcite (up to 0.8 wt.%) may have also slightly suppressed magnetite formation in Exp-350.

Implications for the Hadean H₂-rich hydrothermal environments

It has been considered that the serpentinization of komatiite would have served as the most ubiquitous geological process for hosting H₂-rich hydrothermal environments in the Hadean ocean (e.g., Takai et al. 2006). The present experiments constrained the potential for H₂ generation during the serpentinization of komatiite under CO₂-rich conditions. This work provides further important insights into our interpretation of hatcheries for the emergence and early evolution of life in the Hadean Earth. Exp-350 showed that the fluid H₂ concentration increased up to 2.9 mmol/kg during the serpentinization and carbonation of komatiite. This value is lower than the maximum H₂ concentration of the vent fluids in modern peridotite-hosted hydrothermal systems; the H₂ and ΣCO₂ concentrations in modern peridotite-hosted hydrothermal systems are 12 and 10.1 mM at Logatchev field, 16 and 16 mM at Rainbow field (Charlou et al. 2002), and <1–15 mmol/kg and <0.8 mmol/kg at Lost City field (Kelley et al. 2005), respectively. However, the H₂ concentration in Exp-350 broadly falls within the range of the fluid H₂ concentrations in the Kairei field hosted by both basalt and troctolite (H₂ = 2.5–8.5 mmol/kg and ΣCO₂ < 10.1 mmol/kg; Takai et al. 2004; Gallant and Von Damm 2006; Kumagai et al. 2008; Nakamura et al. 2009). This suggests that in the Hadean, the komatiite still had great potential to generate H₂-rich hydrothermal fluids at high-temperature, even under CO₂-rich conditions. Furthermore, considering that hydrogenotrophic methanogens—long believed to be one of the most probable candidates for the earliest living forms and primary producers on Earth (e.g., Takai et al. 2006; Martin et al. 2008)—indeed dominate in the microbial communities associated with hydrothermal fluids in the Kairei field (Takai et al. 2004), the komatiite-hosted seafloor hydrothermal systems would have fully prepared the energetic basis of hatcheries for the emergence and early evolution of hydrogenotrophic living forms in the Hadean ocean.

In contrast, the fluid H₂ concentration in Exp-250 (0.024 mmol/kg) was much lower than that in Exp-350 and is comparable to the H₂ concentration level in the typical fluids of modern basalt-hosted hydrothermal fields (Charlou et al. 1996, 2000; Gallant and Von Damm 2006). Given that the H₂ generation potential of ultramafic rocks decreases with decreasing temperature within the unstable region of olivine (McCollom and Bach 2009; Klein et al. 2013), the serpentinization of komatiite under CO₂-rich conditions might not have afforded the H₂-rich hydrothermal fluids at temperatures lower than 250 °C. More importantly, carbonate minerals become stable with decreasing temperature in H₂O-CO₂-rock systems (e.g., Shibuya et al. 2013b), which indicates that the carbonate formation and resulting limitation of H₂ generation would occur more significantly at lower temperatures (e.g., <100 °C). Therefore, the possible H₂-rich hydrothermal environments may have been created only in the deep ocean floor because high hydrostatic pressure can elevate the temperature of hydrothermal reaction zones (fluid-rock reaction zones). In other words, other potential candidates, such as the hydrothermal systems in on-land and shallow submarine environments, may not have fully prepared the energetic basis (abundant H₂ availability) of hatcheries for the emergence and early evolution of life in the Hadean Earth due to low-temperature hydrothermal reactions.

Conclusions

High-temperature and high-pressure experiments using komatiite and a CO₂-rich NaCl solution revealed that different hydrothermal reaction temperatures caused differences in the carbonate species in alteration products (iron-rich dolomite at 250 °C and calcite at 350 °C) during the serpentinization of komatiites. The hydrothermal fluid coexisting with dolomite at 250 °C showed high Mg concentrations (up to ca. 40 mmol/kg), which is markedly higher than those in typical modern high-temperature hydrothermal fluids (generally less than 1 mmol/kg). This suggests that, in contrast to modern seafloor hydrothermal systems, the dolomite-bearing komatiite-hosted hydrothermal systems may have served as a source of Mg in the Hadean ocean. The steady-state H₂ concentration in Exp-350 (up to 2.9 mmol/kg) was approximately 100 times higher than that in Exp-250 (up to 0.024 mmol/kg) because the amount of ferrous iron incorporated into carbonate was greater in Exp-250 than in Exp-350. These results perhaps suggest that only high-temperature komatiite-hosted hydrothermal systems have the potential to generate H₂-rich hydrothermal environments, even in the CO₂-abundant Hadean ocean. Further experiments based on a more precise estimation of the ancient atmospheric/oceanic CO₂ levels from geological records will clarify the geochemical nature of komatiite-hosted hydrothermal environments and other potential hatcheries for the emergence and early evolution of life in the Hadean Earth because the ΣCO₂ concentration in the fluid may significantly affect the mineral assemblage of serpentinized komatiites and hydrothermal reactions taking place under various conditions.