Mineral catalyzed organic synthesis in hydrothermal systems: An experimental study using time-of-flight secondary ion mass spectrometry



[1] Hydrothermal fluids enriched in hydrocarbons of apparent abiotic origin vent from Fe-Ni sulfide bearing chimney structures on the seafloor at slow spreading mid-ocean ridges. Here we show results from a hydrothermal experiment using carbon isotope labeling techniques and mineral analytical data that indicate that pentlandite ((Fe2Ni7)S8) enhances formation of C2 and C3 alkanes, while also contributing to the formation of other more complex hydrocarbons, such as alcohols and carboxylic acids. ToF-SIMS data reveal the existence of isotopically anomalous carbon on the pentlandite surface, and thus, for the first time, provide unambiguous evidence that mineral catalyzed surface reactions play a role in carbon reduction schemes under hydrothermal conditions. We hypothesize that hydroxymethylene (-CHOH) serves as intermediary facilitating formation of more complex organic compounds. The experimental results provide an explanation for organic synthesis in ultramafic-hosted hydrothermal systems on earth, and on other water-enriched planetary bodies as well.

1. Introduction

[2] The discovery of high temperature vent fluids at ultramafic-hosted hydrothermal systems, including the Rainbow vent site (36°14′N, Mid-Atlantic Ridge) in particular, revealed the existence of hydrocarbons possibly derived as a consequence of rock-fluid interaction processes at elevated temperatures and pressures [Holm and Charlou, 2001; Charlou et al., 2002; Douville et al., 2002; Charlou et al., 2007]. The temperature of Rainbow vent fluids exceeds 360°C, which suggests still higher temperatures in subseafloor reaction zones. The coexistence of high temperatures and pressures and high concentrations of compositionally diverse hydrocarbons in aqueous fluids is an uncommon association, one that is difficult to predict, in spite of recent advances in thermodynamic data for organic species at elevated temperatures and pressures [Shock and Helgeson, 1990]. One reason for this involves the mechanism of formation of the dissolved organics, which most often has been attributed to Fischer-Tropsch type (FTT) synthesis in which mineral catalyzed surface reactions play a role. Thus, formation of dissolved organics in hydrothermal fluids issuing from Rainbow and similar systems may involve complex kinetic pathways facilitated by constraints imposed by mineral composition and dissolved H2 and CO2 concentrations of the fluid [Berndt et al., 1996; Horita and Berndt, 1999; McCollom and Seewald, 2001, 2003; Foustoukos and Seyfried, 2004; Fu et al., 2007]. Although Rainbow vent fluids are characterized by noteworthy concentrations of dissolved H2 and CO2 [Charlou et al., 2002], the concentration of dissolved H2, in particular, is limited by redox buffering by minerals at elevated temperatures and pressures [Allen and Seyfried, 2003], suggesting that processes other than redox intensity alone can affect hydrocarbon formation.

[3] Rainbow vent fluids also contain dissolved H2S of ∼1 mm/kg [Charlou et al., 2002; Douville et al., 2002]. When these data are considered with constraints imposed by dissolved H2 concentrations (∼16 mm/kg), stability of Fe and Ni-bearing sulfides is indicated [Seyfried et al., 2004]. While the role of sulfur in FTT synthesis remains a subject of debate [Anderson, 1984; Bromfield and Coville, 1999], a number of recent studies [Huber and Wächtershäuser, 1997, 1998; Cody et al., 2000, 2004] have indicated that iron and nickel sulfides may serve as catalysts promoting abiotic organic synthesis. These studies, however, were largely aimed at proceeses related to the origin of life on earth, and can not be extrapolated to interpret unambiguously the formation of hydrocarbons in high temperature vent fluids at Rainbow or in similar hydrothermal systems hosted in ultramafic rocks.

[4] Pentlandite ((Fe, Ni)9S8), together with other Fe-Ni sulfides, is also found in the Allende meteorite [Brearley, 1997, 1999]. These sulfide-bearing minerals coexist with a variety of organic compounds [Botta and Bada, 2002; Sephton, 2002], and hydrous alteration products of ultramafic mineral assemblages, suggesting a causal relationship. Thus, the occurrence of hydrocarbons of an abiotic origin linked to an apparent mineral catalyzed mechanism of formation may not be unique to terrestrial environments [Studier et al., 1972; Mullie and Reisse, 1987].

[5] The association of hydrocarbons in ultramafic-hosted hydrothermal systems on Earth, and in meteorites and the presence of Ni-bearing sulfide minerals, is compelling and needs to be experimentally tested. Accordingly, we conducted an experiment utilizing pentlandite as an FTT catalyst at temperature, pressure, and most importantly, redox conditions that simulate those inferred for the moderately reducing Rainbow ultramafic-hosted hydrothermal system.

2. Experimental Methods

[6] The experiment was carried out in a high-pressure experimental apparatus using a flexible gold-titanium reaction cell [Seyfried et al., 1987]. A powder mixture of equimolar amounts of NiO and FeS (0.01 mol) were loaded with NaCl/HCl fluid, where the total dissolved chloride concentration was 0.50 mol/kg, roughly equivalent to that of seawater. The addition of a small amount of dilute HCl to the starting fluid avoided mineral precipitation during the heat up stage of the experiment, while also precluding carbonate (e.g., siderite) mineral precipitation at experimental conditions, as indicated by results of phase equilibria calculations and analysis of reaction products.

[7] After reaching temperature and pressure conditions of 400°C and 50 MPa, the fluid-mineral system was reacted for three days during which time the mineral components recrystallized into stable phases (see below). Owing to mineral-fluid reactions, pH achieved a value near neutrality at experimental conditions (pH(P,T) ≈ 5), which is very near that estimated for the Rainbow vent fluids [Allen and Seyfried, 2003]. Background sources of carbon that were associated with the mineral components (NiO, FeS) dissolved, as indicated by the chemical composition of a fluid sample taken from the reactor at this time. Subsequent to the three-day “pretreatment” interval, a solution of 13C-labeled formic acid (99% H13COOH) was added to the reaction cell by a high-pressure fluid delivery system. Thus, the addition of the isotopically labeled dissolved carbon represents the starting point (time-zero) against which subsequent changes in fluid chemistry are compared.

[8] Scanning Electron Microscopy (SEM) and X-ray Diffraction (XRD) analysis of mineral products confirmed that pentlandite ((Fe2Ni7)S8) was the dominant alteration mineral, although minor magnetite (Fe3O4) and heazlewoodite (Ni3S2) were also observed (see auxiliary materials).

3. Results and Discussion

[9] Owing to the relatively high temperature of the experiment, formic acid, which served as the source of isotopically labeled dissolved carbon, can be expected to decarboxylate quantitatively [Yu and Savage, 1998]. Based on the concentration and mass of formic acid added to the reaction cell, therefore, dissolved concentrations of CO2 and H2 of 40 mmol/kg each can be predicted. These concentrations, however, exceed those actually measured in the first sample at 50 hours by approximately 23 and 68%, respectively (Table 1). The less than predicted dissolved concentrations of both species suggest hydrocarbon formation. It is unlikely that the change in dissolved H2 is entirely related to hydrocarbon formation, since recrystallization of component reactants to pentlandite plus additional minor secondary minerals can be shown to be an H2-consuming and ultimately redox buffering process. Indeed, at the temperature, pressure and bulk composition of the experiment, phase equilibria constraints indicate an invariant system, although the lack of data for solid solution effects on the thermodynamic properties of Fe-Ni sulfide minerals makes this difficult to confirm unambiguously. The steady state concentrations of dissolved H2 and H2S during the nearly 1000-hour experiment (Table 1), however, are consistent with phase equilibria control. Moreover, that dissolved H2 and H2S concentrations measured during the experiment are virtually identical to analogous data for Rainbow vent fluids [Charlou et al., 2002] allows the possibility of similar phase relationships in ultramafic-hosted hydrothermal systems as well [Charlou et al., 2007] (Figure 1).

Figure 1.

Activity-activity diagram depicting redox phase equilibria for the NiO-H2S-H2O-HCl (solid lines) and FeO-Fe2O3-H2S-H2O-HCl (dashed lines) systems at 400°C, 50 MPa. Pentlandite ((Fe2Ni7)S8) and magnetite (Fe3O4) are predicted to be stable phases at experimental conditions, which is also likely for subseafloor reaction zones at Rainbow (36°14′N, MAR). The existence of heazlewoodite (Ni3S2) solid solution and lack of corresponding thermodynamic data may account for the difference between predicted mineral assemblage and that actually observed following the experiment (see text). Data for H2(aq) and H2S(aq) for Rainbow vent fluids are from Charlou et al. [2002]. Thermodynamic calculations were performed taking explicit account of upgrades to the SUPCRT92 database (see Seyfried et al. [2004]). Thermodynamic properties of pentlandite solid solution are from Warner et al. [1996] and calculated using the approach outlined by Helgeson et al. [1978].

Table 1. Dissolved Concentrations of Selected Aqueous Species During the Experiment and Reported for the Rainbow Hydrothermal Systema
Time, hH2, mH2S, mΣ12CO2,b mΣ13CO2, mCH4(16),c mCH4(17), μC2H6(30), μC2H6(31), μC3H8(44), μC3H8(45), μC3H8(46), μC3H8(47), μ
  • a

    Concentrations are in mmol/kg (m) or μmol/kg (μ). Analytical errors are estimated to be ≤5%. See Charlou et al. [2002].

  • b

    ΣCO2 is the total dissolved carbonate species, dominated by aqueous CO2 (>95%) as in situ pH value is ∼5.

  • c

    The number in each parenthesis indicates the mass-to-charge ratio (m/z) for each aqueous alkane species. Ethane with m/z = 32 cannot be identified because of the increased background level due to the existence of O2.

  • d

    Time zero was set after heating for 3 days at 400°C, 50 MPa, which served as a pretreatment period during which time pentlandite formed and background sources of carbon dissolved (see text).

0d31.90.5 0.06       
5Injected H13COOH in NaCl and HCl solution 
       H2, mH2S, mCO2, mCH4, mC2H6, μC3H8, μ
Rainbow hydrothermal system161.2162.51.10.05

[10] Isotopically pure 13C methane and other alkane species (C2H6 and C3H8) were detected in all fluid samples (Table 1). In agreement with previous experimental observations [Foustoukos and Seyfried, 2004], 13C was also incorporated with 12C into molecular structures to form combined 12C and 13C-bearing isotopomers, such as C2H6 (m/z = 31) and C3H8 (m/z = 45, 46). Overall, the concentrations of isotopically labeled dissolved alkanes increased during the first 650 hours, and then remained relatively constant for the remainder of the experiment indicating attainment of steady state conditions, perhaps induced by the gradual loss of effectiveness of the pentlandite catalyst in the closed reactor. That substantial changes in dissolved CO2 and H2 actually preceded alkane formation, however, suggests that formation of other presently unidentified hydrocarbons may have contributed to alkane formation, as recently proposed by Fu et al. [2007]. The percentages of alkanes of unambiguous abiotic origin (entirely with 13C labeled) are ∼8% for methane and ∼2% for propane. In comparison with previous experiments using different mineral catalysts [Foustoukos and Seyfried, 2004], the concentration of isotopically pure 13C propane by the end of this experiment (0.2 μmol/kg) is significant, especially when allowance is made for differences in dissolved H2 concentrations. Moreover, abiotic propane from the present experiment is approximately a factor of four greater than that for Rainbow vent fluids [Charlou et al., 2002], underscoring the effectiveness of pentlandite as a catalyst for formation of longer chain alkanes. A sequence of reactions with different mineral catalysts along the fluid flow path might contribute to the much higher concentration of dissolved methane in Rainbow fluids than observed for the experiment [Fu et al., 2007].

[11] X-ray Photoelectron Spectroscopy (XPS) data showed that the two components (NiO and FeS) used to form the reaction product during the experiment contained carbon compounds, which correspond to functional groups containing C-C or C-H bonds (285.0 eV), alcohols (286.6 eV), and Ni(CO)4 or Fe(CO)5 (289.5 eV), respectively (see auxiliary materials). These components were likely derived from the manufacturing process. The C 1s spectra for the pentlandite-dominated product minerals, however, reveal significant enrichment of surface carbon, confirming uptake from solution, but also the presence of carbon compounds indistinguishable from reactant sources. The primary peak of product carbon at 285.0 eV and the smaller peak at 286.6 eV, indicate the presences of alkyl (CH3-CH2-) and methylene (-CH2-) groups, or other basic hydrocarbon units containing C-C or C-H bonds, and alcohol groups (C-OH), respectively. Moreover, an entirely new surface carbon phase with a peak at 288.9 eV, associated only with product minerals that formed during the experiment, is best attributed to carboxyl groups (-COOH). Neither carbide (283.6 eV or 283.9 eV) nor graphite (284.5 eV) was observed associated with the mineral products (see auxiliary materials).

[12] The potential existence of background sources of carbon on mineral reactants underscores the need to use isotopically anomalous carbon to monitor carbon reduction processes, as has been emphasized previously [McCollom and Seewald, 2001; Foustoukos and Seyfried, 2004]. Here we applied this approach to both the fluid and coexisting minerals. To identify the 13C source of carbon-bearing species on mineral surfaces, while simultaneously distinguishing between 13C (amu = 13.0034) and 12C1H (amu = 13.0078), which have the same nominal mass, Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) analysis was performed. ToF-SIMS analysis of the starting mineral component surfaces (Figure 2) revealed the dominating presence of 12C, with trivial amounts of 13C. The opposite is true, however, of reaction products, which show substantial enrichment of 13C-bearing species, although distinctly lesser amounts of 12C1H are still apparent. Thus, these data confirm that the metastable carbon species on the surface of reaction products that were identified by XPS were abiotically derived by CO2 reduction under hydrothermal conditions, demonstrating the unambiguous linkage between mineral surface catalysis processes and hydrocarbon formation.

Figure 2.

High mass resolution ToF-SIMS spectra of 13C (amu = 13.0034) and 12CH (amu = 13.0078) obtained from reactants (NiO and FeS) and pentlandite-dominated reaction product. Both negative and positive ions were collected, although only negative ion results are shown here, because of higher intensity (104) of these species than that of positive ions (101). The noteworthy abundance of 12C1H on NiO and FeS confirms the existence of “background” carbon, which is supported by fluid chemical data measured after the 3-day preheating stage. 13C-bearing species, however, dominate the isotopic composition of carbon on mineral surface of pentlandite. These data confirm the role of mineral surface reactions in carbon reduction schemes at elevated temperatures and pressures in keeping with XPS data.

[13] Even though Fischer-Tropsch type synthesis has a long history in industry, the reaction mechanism is still not completely understood. Carbide [Brady and Pettit, 1981], hydroxymethylene (-CHOH) [Storch et al., 1951; Ekstrom and Lapszewicz, 1987], and CO-insertion [Pichler and Schultz, 1970] are just three of the more common mechanisms that have been proposed. The carbide mechanism is decidedly more common and effective, but requires unusually reducing conditions often characterized by high concentrations of CO and H2 in a coexisting gas phase [Brady and Pettit, 1981]. Aqueous fluids at relatively high temperatures and pressures in subseafloor hydrothermal systems, however, likely preclude FTT synthesis by the carbide mechanism constraints imposed by coexisting mineral assemblages, which limit reducing conditions (Figure 1). Thus, in mafic and ultramafic-hosted hydrothermal systems and their experimental analogues, the hydroxymethylene mechanism is more likely. This mechanism is generally consistent with XPS data, which indicate formation of oxygenate compounds including alcohols and carboxylic acids on the surface of mineral reaction products.

[14] A scenario to describe FTT synthesis during the present experiment might include the following: CO and H2 adsorption on the catalyst surface, with subsequent conversion to hydroxymethylene. C-C bonds can be formed through a condensation reaction involving loss of H2O between hydroxymethylene groups [Storch et al., 1951; Kummer and Emmett, 1953]. Chain growth continues at mineral surface by incorporation of additional hydroxymethylene groups. Dissolved H2 can serve as the chain terminator to break C-C bonds, yielding hydrocarbons or alcohols, which is an endothermic process and thermodynamically favored at high temperature. Partial oxidation of hydroxymethylene can result into formation of carboxylic acids and other oxygenates.

4. Conclusions

[15] We have demonstrated hydrothermal synthesis of dissolved alkane species and oxygen-bearing organic compounds on the surface of pentlandite, which served as a catalyst for carbon reduction reactions. The need for only moderately reducing conditions for synthesis of diverse and abundant dissolved organic compounds is highly significant in that similar redox constraints are likely in ultramafic-hosted hydrothermal systems where hydrocarbons in hydrothermal vents fluids have recently been recognized. Identification of metastable reaction products on the mineral catalyst (pentlandite) surface by XPS analysis along with ToF-SIMS measurements confirms the previously inferred role of mineral surfaces as catalytic agents for carbon reduction processes. That pentlandite is not only a likely product of hydrothermal alteration of ultramafic rocks at mid-ocean ridges, but also observed in association with organic compounds in meteorites, broadens significantly the implications of this research.


[16] We thank Rick Haasch and Tim Spila at Center for Microanalysis of Materials, University of Illinois, for assistance in XPS and ToF-SIMS analysis. We would also like to thank George Cody (CIW) for his comments and suggestions, which greatly improved the manuscript. This work was supported by NSF OCE-0549457 and ACS PRF-41885-AC2. Helpful comments by Jean Luc Charlou and an anonymous reviewer are appreciated.