Report

# Hydrocarbons in Hydrothermal Vent Fluids: The Role of Chromium-Bearing Catalysts

See allHide authors and affiliations

Science  14 May 2004:
Vol. 304, Issue 5673, pp. 1002-1005
DOI: 10.1126/science.1096033

## Abstract

Fischer-Tropsch type (FTT) synthesis has long been proposed to account for the existence of hydrocarbons in hydrothermal fluids. We show that iron- and chromium-bearing minerals catalyze the abiotic formation of hydrocarbons. In addition to production of methane (CH4aq), we report abiotic generation of ethane (C2H6aq) and propane (C3H8aq) by mineral-catalyzed hydrothermal reactions at 390°Cand 400 bars. Results suggest that the chromium component in ultramafic rocks could be an important factor for FTT synthesis during water-rock interaction in mid-ocean ridge hydrothermal systems. This in turn could help to support microbial communities now recognized in the subsurface at deep-sea vents.

Vent fluids issuing from ultramafic-hosted hydrothermal systems at mid-ocean ridges not only contain abundant methane but are also enriched in propane, ethane, and many other dissolved hydrocarbons (1, 2). It is likely that the occurrence and distribution of these hydrocarbons is the result of FT T synthesis, where oxidized forms of dissolved carbon are reduced to hydrocarbons by reaction with H2aq. In general, this process can be described schematically as follows: $Math$ $Math$(1) The formation and distribution of alkanes produced in hydrothermal experiments at elevated pressure and temperature suggest that the reactions are catalyzed by minerals (3). As such, the chemical and physical properties of the catalyst play a key role in hydrocarbon yield. For example, formation of relatively small amounts of methane was reported in experiments involving reaction of CO2-bearing aqueous fluid with different minerals (hematite, magnetite, olivine, serpentine, and Ni-Fe alloy) (4, 5). The Ni-Fe alloy (awaruite), in particular, appears to be an excellent catalyst for CO2aq conversion to CH4aq (6). Although abiotic methane was inorganically generated during these experiments, no other alkanes were produced. The relative lack of hydrocarbons other than methane, however, brings into question an origin by FTT synthesis of the complex hydrocarbons in vent fluids issuing from ultramafic-hosted hydrothermal systems (7). McCollom and Seewald (5) speculated that it is only in the presence of a discrete gas phase that abiotic synthesis of the more complex hydrocarbon species may be at all possible. Here, we report results of a hydrothermal experiment indicating that FeCr oxide (e.g., chromite) is a catalyst for FTT synthesis of longer chain hydrocarbons. The chromium content of fresh oceanic ultramafic rocks is nearly 3000 ppm (810) and is preferentially concentrated in orthopyroxene (11, 12), a particularly reactive mineral in ultramafic rocks (3, 13). Orthopyroxene alteration can be expected to provide Cr for chromite, a common accessory mineral, especially in enstatite-rich peridotite or bastite.

Our experiments were performed at 390°C and 400 bars, conditions that approximate those inferred for ultramafic-hosted hydrothermal alteration at Rainbow (36°N) and Logatchev (14°N) on the Mid-Atlantic Ridge (1, 2, 7). In addition to abundant hydrocarbons, vent fluids from these hydrothermal systems have substantial amounts of dissolved H2aq (2). Reducing conditions undoubtedly result from the hydrolysis of olivine, or more likely orthopyroxene, giving rise to the formation of magnetite together with talc and/or serpentine (3, 13).

Experiments were conducted in a flexible gold-cell hydrothermal apparatus (14), which allows fluid sampling at experimental conditions while also permitting introduction of fluid reactants (15). An added advantage of the gold-titanium reaction cell is its inherent lack of catalytic activity. Therefore, high dissolved concentrations of CO2 and H2 can coexist for long intervals at temperatures and pressures as high as 400°C and 500 bars without generation of appreciable amounts of reduced carbon species (16). Thus, in the absence of appropriate mineral catalysts, generation of reduced carbon species is inhibited.

To trace carbon sources and sinks during the experiment, we added 13C-enriched NaHCO3 (∼99% 13C) to the fluid. The starting fluid also contained NaCl (0.56 mol/kg) to approximate the bulk chemistry of axial vent fluids. Moreover, to facilitate chromite formation under highly reducing conditions (17), FeO was combined with Cr2O3 and the 13C-bearing aqueous fluid (18). The existence of FeO in amounts greater than needed to form stoichiometric chromite permitted the formation of magnetite and produced high H2aq concentrations (19). Time-series observations of fluid samples, however, indicated sluggish FeO hydrolysis during the early stages of the experiment (Table 1). This was likely because the pH(T,P) (pH at elevated temperature and pressure) was in excess of neutrality as a result of the dissolved NaHCO3 in the starting fluid. To lower pH(T,P) to 4.7 to 4.8, in keeping with natural vent fluids from ultramafic-hosted hydrothermal systems (13), we injected a small amount of dilute HCl into the contents of the reaction cell. Dissolved H2aq immediately increased (Table 1).

Table 1.

Abundances of dissolved gases in fluids coexisting with FeO-Cr2O3 (experiment 1) and FeO (experiment 2) at 390°C and 400 bars. Concentrations of the 13C species were corrected for hydrogen radical losses and for predicted natural carbon isotope abundances derived from background carbon sources. Analytical errors are estimated to be ≤5%. ΣFormate corresponds to measured HCOO-, HCOOHaq, and NaHCOOaq; Σ13CO2aq refers to 13CO2aq, H13CO3-, and 13CO32-. Estimates of pH(T,P) are based on constraints imposed by iron solubility (15) and NaHCO3-NaCl-H2O-magnetite-fluid equilibria (17).

Time (hours) H2aq (mmol/kg) Σ13CO2aq (mmol/kg) Σ12CO2aq (mmol/kg) 13CH4aq (μmol/kg) 13C3H8aq (μmol/kg) 13CH4aq/13CO2aq (%) ΣFormateView inline (mmol/kg) 12C2H6aq (μmol/kg) 12C3H8aq (μmol/kg) pH(T,P)
Experiment 1: FeO-Cr2O3
0 30
126 41 22 1.67 2 0.01 1.9 8.7
126 Injection of 2.5 N HCl (2 ml)
534 219 26 1.37 63 0.24 0.2
774 220 26 1.41 85 0.3 0.32 0.2 55 49 4.8
1062 192 24 1.28 122 1.0 0.50 0.2 55 57 4.8
Experiment 2: FeO
0 30
48 19 24 1.05 1 0.01 1.0
264 48 25 1.05 3 0.01 2.8 8.8
288 Injection of ∼330 mmol/kg H2aq (5 ml)
384 81 21 0.92 5 0.02 4.4 8.8
408 Injection of ∼390 mmol/kg H2aq (3.5 ml)
600 106 23 1.00 16 0.07 5.4 8.7
768 115 22 1.15 22 0.10 5.6 8.7
1152 108 20 0.84 39 0.19 4.1 8.6
1752 117 20 0.87 89 0.1 0.45 6.3 29 24 8.8
2880 121 22 0.94 195 0.3 0.90 5.9 52 36 8.7
• View inline* Concentrations of the dissolved C2-C3 carboxylic acids are below detection limits (10 μmol/kg).

• Elevated concentrations of isotopically labeled methane (13CH4aq) were generated during the experiment, confirming FTT synthesis. After 534 hours, for example, 13CH4aq production reached 63 μmol/kg (Table 1); by the end of the experiment, this increased to 122 μmol/kg. This is equivalent to 13CO2aq conversion of 0.5%, which, when taking account of reaction time, is higher than the results of previous isotopically controlled experiments that were run at lower temperatures but lacked chromite as a discrete phase (4, 5).

FeO- and Cr2O3-bearing systems also affected hydrocarbon chain growth mechanisms (20). After 774 hours of reaction, propane of abiotic origin (13C3H8aq) was observed at a concentration of 0.3 μmol/kg. The isotopically pure 13C propane fraction (mass/charge ratio m/z = 47) increased with time to 1.0 μmol/kg by the end of the experiment (Fig. 1A). Owing to the presence of oxygen (m/z = 32) during sample processing, it was not possible to confirm unambiguously the likely existence of abiotic ethane (13C2H6aq). We did, however, observe anomalous amounts of m/z 31 (13C12CH6aq) (relative to that predicted from natural carbon abundances), which is consistent with abiotic ethane addition of ∼4 μmol/kg to background sources of ethane, with a slight increase in abundance with reaction progress (Fig. 1A). Background ethane and other trace hydrocarbons are likely by-products of the small amounts of organic contaminants associated with both FeO and Cr2O3 reactants; this underscores the need to use isotopic label techniques, as has been emphasized elsewhere (4, 5).

The reaction path processes we observed are not consistent with simple probability models (21). The yield rate (% C3H813CO2aq) of propane with m/z = 45 was 8 to 17 times that of isotopomers 46 and 47 (Fig. 1A). This result suggests that 12C sources in the reactants interacted with fluid-sourced 13C to form three-carbon chains during hydrothermally induced FTT synthesis. Thus, preferential chain growth by reaction of residual alkyl (12CH3-12CH2-) groups on mineral surfaces with methylene (13CH2), which is derived from hydrogenation and adsorption of isotopically labeled carbon, could account for the most abundant 13C-bearing propane isotopomer (m/z = 45). X-ray photoelectron spectroscopy studies of hydrothermally altered C-bearing magnetite grains provide evidence for the existence of residual alkyl groups, even after reaction at temperatures as high as 400°C (16). Formation of the more 13C-enriched isotopomers (m/z = 46, 47) requires construction of alkyl groups together with methylene adsorption, consistent with models (22). Accordingly, the relatively low concentrations of these isotopomers are to be expected. Using the same procedure outlined earlier for ethane, we can infer from the distribution of propane isotopomers that as much as 12 μmol/kg of abiotic propane formed during the experiment, increasing in amount with increasing time (Fig. 1A). Restricting propane to the entirely inorganic derivative (m/z = 47), the concentration is still approximately an order of magnitude greater than that reported for vent fluids issuing from the ultramafic-hosted Rainbow hydrothermal system (2).

Recent experimental data (23, 24) suggest reversible equilibria between n-alkanes and carboxylic acids of corresponding chain length n, whereas lab and field data indicate that carboxyl carbon in carboxylic acid is susceptible to isotopic exchange, especially at high-pH conditions (25, 26). Thus, by this mechanism, alkanes derived from background sources could be incrementally enriched in 13C and would not involve FTT synthesis at all (27). To test this idea, we performed an additional experiment that involved only FeO as a catalyst. Temperature, pressure, composition of the aqueous phase other than pH (see below), and ever-present background carbon sources were identical to those of the Cr2O3-bearing experiment. To optimize isotope exchange rates and enhance the abundance of carboxylic acid and acid anions, we conducted the experiment at a pH(T,P) of ∼8.8 (Table 1). This experiment, therefore, was specifically designed to maximize the effectiveness of the isotope exchange model.

Although substantial isotopically anomalous methane was generated during the FeO experiment (Table 1), this was not the case for 13C3H8aq, which reached a concentration of only 0.3 μmol/kg even after 2880 hours of reaction—about three times the reaction period for the FeO-Cr2O3 experiment (Fig. 2). Moreover, a relatively low concentration and near-normal 13C abundance of ethane was obtained in the pure FeO experiment. Taken together, these findings suggest limited effectiveness of the carboxyl carbon (carboxylic acid)–alkane isotope exchange mechanism, high pH notwithstanding. Apparently, at 390°C, C2-C3 carboxylic acids and acid anion concentrations are simply too low for the isotope exchange mechanism to be effective. Thus, it is the catalytic properties of the Cr-bearing systems that account best for the formation of low molecular weight hydrocarbons by FTT synthesis.

Our experimental results show that Cr2O3, in combination with Fe-bearing oxides, is an effective catalyst for FTT synthesis of C1,C2, and C3 hydrocarbons. Indeed, supported chromium oxide catalysts have been extensively used for hydrogenation and dehydrogenation reactions in gas-phase applications (28). The catalytically active sites have been inferred to consist of a Cr3+ cation and an oxygen anion, where the Cr3+ cation is situated next to a surface anion vacancy (29). Although we have emphasized the effectiveness of chromium and iron oxides to catalyze methane, ethane, and propane synthesis under hydrothermal conditions, it is likely that the catalytic properties of still other minerals are required to account for the full range of linear long-chain hydrocarbons reported in vent fluids issuing from the ultramafic-hosted Rainbow hydrothermal system (7). Hydrocarbon chain growth can also be facilitated by the presence of residual alkyl groups on mineral surfaces, as suggested from our experimental results. This, together with the recognized existence of residual hydrocarbons in mantle rocks (30), may play a key role, both quantitatively and qualitatively, in the distribution and isotopic composition of hydrocarbons in hydrothermal vent fluids. Mineral-catalyzed production of hydrocarbons in ultramafic-hosted hydrothermal systems could also help to account for the diverse communities of Archaea and Eubacteria inhabiting chimney deposits at vents, both in the present (31) and likely in the geologic past (32).

Supporting Online Material

Materials and Methods

References

## References

View Abstract