Hydrothermal synthesis of long-chain hydrocarbons up to C₂₄ with NaHCO₃-assisted stabilizing cobalt

Significance

The mechanisms by which abiotic, long-chain hydrocarbons are produced in natural alkaline hydrothermal systems are unknown, as only short-chain hydrocarbons (<5 carbons) have been experimentally observed to date. Here, we demonstrate how the hydrothermal reduction of bicarbonate into long-chain hydrocarbons (≤24 carbons) occurs through the use of iron and cobalt metals. In contrast to the traditional Fischer–Tropsch synthesis, in which water is the driving force for catalyst deactivation, Co exhibits unique catalytic stability in hydrothermal conditions through bicarbonate-assisted CoO_x reduction, thus promoting the carbon−carbon coupling process with the synergistic effect from the iron hydroxyl group. This finding helps to explain the abiogenic origin of petroleum, life’s emergence, and further contributes to artificial carbon dioxide utilization in the chemical industry.

Abstract

Abiotic CO₂ reduction on transition metal minerals has been proposed to account for the synthesis of organic compounds in alkaline hydrothermal systems, but this reaction lacks experimental support, as only short-chain hydrocarbons (<C₅) have been synthesized in artificial simulation. This presents a question: What particular hydrothermal conditions favor long-chain hydrocarbon synthesis? Here, we demonstrate the hydrothermal bicarbonate reduction at ∼300 °C and 30 MPa into long-chain hydrocarbons using iron (Fe) and cobalt (Co) metals as catalysts. We found the Co⁰ promoter responsible for synthesizing long-chain hydrocarbons to be extraordinarily stable when coupled with Fe−OH formation. Under these hydrothermal conditions, the traditional water-induced deactivation of Co is inhibited by bicarbonate-assisted CoO_x reduction, leading to honeycomb-native Co nanosheets generated in situ as a new motif. The Fe−OH formation, confirmed by operando infrared spectroscopy, enhances CO adsorption on Co, thereby favoring further reduction to long-chain hydrocarbons (up to C₂₄). These results not only advance theories for an abiogenic origin for some petroleum accumulations and the hydrothermal hypothesis of the emergence of life but also introduce an approach for synthesizing long-chain hydrocarbons by nonnoble metal catalysts for artificial CO₂ utilization.

Apart from the well-known microbial and thermogenic origin of hydrocarbons, the scenario of their abiogenic origin has emerged as an assumption and has been demonstrated on a few occasions (1, 2). Abiotic hydrocarbons production from inorganic carbon is considered to be driven by serpentinization in H₂-rich alkaline hydrothermal fluids, and, recently, the hydrocarbons generated are proposed to be transported through deep faults to shallower regions in the Earth’s crust and contribute to some petroleum reserves (3). In addition, the current scenarios suggest that the abiotic synthesis of hydrocarbons in natural hydrothermal systems is also a potential source of prebiotic organic compounds for life’s emergence (4). However, the mechanisms by which long-chain hydrocarbons are synthesized remains unclear and controversial because such reactions presently lack experimental support despite numerous artificial simulations of hydrocarbon formation, although short-chain hydrocarbons (<C₅) have been synthesized (5).

In general, the abiotic formation of hydrocarbons is assumed to proceed through the Fischer–Tropsch synthesis (FTS) pathway catalyzed by natural minerals (6, 7). The predominant, oxidized forms of dissolved carbon under high pressure, CO_2(aq) or HCO₃⁻ (>70%), react with dissolved hydrogen H_2(aq) to produce hydrocarbons. The process can be described as follows:

$${\text{CO}}_{2\left(\text{aq}\right)}+[3+(1/\text{n})]{\text{H}}_{2\left(\text{aq}\right)}\text{ → }(1/\text{n}){\text{C}}_{\text{n}}{\text{H}}_{\text{2n}+2}+{\text{2H}}_2\text{O,}$$ [1]

$${\text{HCO}}_3{}^{-}+[3+(1/\text{n})]{\text{H}}_{2\left(\text{aq}\right)}\text{ → }(1/\text{n}){\text{C}}_{\text{n}}{\text{H}}_{\text{2n}+2}+{\text{2H}}_2\text{O}+{\text{OH}}^{-}.$$ [2]

Fe(II) in the rock-forming minerals reduces H₂O to H₂, while the Fe(II) is converted to Fe(III) (8). In addition to Fe(II), as the H₂ electron donor, clues from biological mechanism and the chemical literature suggest that the native metals can act as the prebiotic source of electrons to fix CO₂ (9–13). The native Fe or the nickel–iron (awaruite, Ni₃Fe) and cobalt–iron (wairauite, CoFe; CoFe₂) alloys have been found in the Earth’s serpentinized ultramafic rocks, where the occurrence of the cobalt–iron alloy is associated with iron minerals, such as chromite and awaruite (14–18). Thus, the native Fe, abundant natural Fe species, or its minerals combined with other transition metals are considered as promising reductants and/or catalysts to reduce CO₂ in hydrothermal systems were thermodynamic conditions to be far enough from equilibrium. To prove the hydrocarbon formation under geochemical conditions, many efforts have been made, and only short-carbon chains (≤C₅) can be produced from CO₂/HCO₃⁻/CO₃⁻, for example, with Fe–Cr oxide (chromite) (7) or awaruite (Ni₃Fe) in alkaline conditions (200 to ∼400 °C, 40 to 50 MPa) (1) or with Fe-bearing dolomite (ankerite) at a higher temperature and pressure (600 to ∼1,200 °C and 1 to ∼6 GPa) (19). However, the long-chain hydrocarbons (>C₅) have not been formed, and it remains an open question as to how hydrothermal conditions might favor direct, long-chain hydrocarbon synthesis from CO₂/HCO₃⁻/CO₃⁻. Indeed, the lack of direct experimental evidence and mechanisms for long-chain hydrocarbon formation from CO₂/HCO₃⁻ in artificial alkaline hydrothermal systems has led to serious doubts regarding the supposed abiogenic origin of hydrocarbons, as well as the hypothesized evolutionary transition from geochemical to biochemical processes. Identifying hydrothermal settings that activate H₂–CO₂ redox reactions with a transition metal catalyst for long-chain hydrocarbon synthesis is an important step toward understanding prebiotic chemistry on the early Earth.

Additionally, the utilization of CO₂ as a feedstock for chemical fuel synthesis has received considerable research interest to reduce the carbon footprint and achieve a carbon-neutral cycle (20–22). Our previous efforts of hydrothermal CO₂ reduction with metals have shown that CO₂/HCO₃⁻ can be easily converted into C₁ products such as formate and methane (23, 24); however, it remains challenging to synthesize products beyond C₃, especially of long-chain hydrocarbons. In the traditional FTS performed with dry CO₂ gas and H₂, Co does exhibit catalytic activity for synthesizing long-chain hydrocarbons, yet it must be born in mind that Co is extremely unstable because of its easy oxidation when exposed to wet vapor (25–27). Moreover, the CoO_x formed is difficult to reduce, as the ambient water acts as an oxidant, thus deactivating the Co catalyst (28). Thus, this chemical engineering field has been left to the noble metal catalysts (29–31) or organic solvents (32), seriously limiting their applications. Recently, we found that a minor portion of the Co remained in the metallic phase during the hydrothermal reduction of CO₂ to acetate (33). Although the mechanism by which the portion of the Co phase remains under hydrothermal conditions is unknown, the phenomenon brought to mind the unique role of hydrothermal systems to stabilize the valence state of Co, thus making possible the synthesis of long-chain hydrocarbons.

Here, we report the hydrothermal synthesis of long-chain hydrocarbons with the use of metal Fe (as a reductant) and Co as a catalyst by designing a dynamic redox equilibrium system for maintaining Co stability. We demonstrate that CoO_x can be reduced via CoCO₃ in the presence of bicarbonate under hydrothermal conditions that inhibit Co deactivation in water, as honeycomb nanosheets are generated in situ as new motif. The unique stability and honeycomb nanosheets of Co coupled with the synergistic effect of Fe−OH and Co metal result in long-chain hydrocarbon formation (up to 24 carbons) from HCO₃⁻ reduction at ∼300 °C and ∼30 MPa. Such hydrothermal synthesis in a manner comparable to geochemical expectations of serpentinization of ultramafic rocks comprising the oceanic crust is of significance not only to the abiogenic origin of petroleum, the emergence of life, but also to the demonstration of how such hydrocarbons can be generated in the absence of noble metals of significance to the chemical industry.

Results and Discussion

NaHCO₃-Assisted Stabilization of Co.

Although the existence of a minor Co metal phase was observed in our past research into the hydrothermal reduction of CO₂ to acetate, the mechanisms by which a portion of Co⁰ was retained and stabilized in water remained unknown. Considering the actual oxidation of Co⁰ in water, we surmise that the stability could be attributed to a dynamic equilibrium between oxidation and reduction during the hydrothermal reactions. The direct reduction of CoO_x under low temperature, such as 300 °C, is difficult. Thus, we first investigated whether, and how, CoO_x could be reduced under hydrothermal conditions. The alkaline hydrothermal environment involves the participation of dissolved carbon compounds, which may react with CoO_x and change the reaction pathway. Here, we conducted CoO reduction at 300 °C with H₂ gas as a function of time in alkaline hydrothermal conditions. The NaHCO₃ is selected as the carbon source to create alkaline conditions (pH ∼8.23) (SI Appendix, Table S1), as it has been considered the dominant species among all the dissolved carbon species [CO_2(aq), H₂CO₃, HCO₃⁻, and CO₃²⁻] in CO₂-saturated alkaline vent fluids (pH ∼9) (1, 8, 34). In the presence of HCO₃^–, the CoO was quickly reduced to Co within 1 h (Fig. 1A), and a small amount of CoCO₃ was detected at a short reaction time of 0.5 h and a low temperature of 200 °C (Fig. 1 A and B). These results indicate that CoO can be reduced to Co⁰ via a CoCO₃ intermediate. In contrast, in the absence of HCO₃^–, Co(OH)₂ was generated, and only a very small quantity of CoO was converted to Co. Thermodynamic calculation also demonstrated that the reduction of CoO via CoCO₃ intermediate is more favorable (Fig. 1C). Taken together, these results demonstrate that the presence of HCO₃^– not only promotes the reduction of CoO into Co⁰ but, thereafter, keeps it stable under these hydrothermal, reducing conditions.

Fig. 1.

NaHCO₃-assisted, hydrothermal CoO reduction. (A) Time-dependent XRD patterns of solid sample after hydrothermal reduction of CoO by H₂ at 300 °C in the presence (red lines) or absence (blue lines) of NaHCO₃. (B) Temperature-dependent XRD patterns of solid sample after hydrothermal reduction of CoO by H₂ for 3 h in the presence of NaHCO₃. Experimental conditions are the following: H₂, 5 MPa; NaHCO₃, 80 mmol; CoO, 40 mmol; water filling, 50%; pH ∼8. The vertical black lines in A and B are the XRD peaks of standard Co(OH)₂ (JCPDS No. 30–0443), Co (JCPDS No. 01–1278), and CoCO₃ (JCPDS No. 11–0692). (C) Gibbs energy of CoO reduction via Co(OH)₂ and CoCO₃ pathway, respectively. JCPDS: Joint Committee on Powder Diffraction Standards.

Hydrothermal Synthesis of Long-Chain Hydrocarbons by Co.

After demonstrating the stability of metallic Co under hydrothermal conditions, we tested whether long-chain hydrocarbons could be synthesized by using NaHCO₃ as a carbon source, Fe as a reductant, and Co⁰ as a catalyst. The reduction of NaHCO₃ was executed in a 42-mL stainless steel autoclave with a water-filling level of 50% at 200 to 300 °C and 20 to 30 MPa—approximating a natural hydrothermal environment (temperature of ≤400 °C and pressure of ≤100 MPa) (35–38). A control experiment was performed first using only Fe without Co, whereby only HCOO⁻ and minor methane were obtained (SI Appendix, Table S2 and Fig. S1 and Fig. 2A). To our delight, the formation of long-chain hydrocarbons was clearly observed in the presence of metallic Co (Fig. 2), suggesting that Co⁰ exhibited catalytic activity for C−C coupling. Moreover, increases in the concentration of NaHCO₃, as well as of the operating temperature of the system, led to a continually increasing hydrocarbon number, as to be expected of hydrocarbon production kinetics (SI Appendix, Fig. S2). Also, the high-NaHCO₃ concentration is found to enhance the oxidation of Fe (SI Appendix, Fig. S3), leading to the accumulation of in situ hydrogen and thereby concomitant high pressure forcing long-chain hydrocarbon formation. In contrast, on reducing the concentration of Fe and NaHCO₃ by half, the amount of short-chain hydrocarbons decreased, and the formation of long-chain hydrocarbons was not observed (SI Appendix, Table S2).

Fig. 2.

Identification of hydrocarbons synthesized by the hydrothermal reduction of NaHCO₃. (A) Total ion chromatogram from GC-MS analysis of liquid products synthesized in the presence (red line) or absence (blue line) of Co. Experimental conditions are the following: Fe, 80 mmol; NaHCO₃, 80 mmol; Co, 40 mmol; water filling, 50%; 300 °C, 9 h, pH 8.23, 30 MPa. The number, n, in the C_n label indicates the carbon chain length of the linear alkanes (only even-numbered peaks are labeled). (B) Enlarged spectra of alkenes and branched alkanes of C₁₂ compounds marked with the red square in A. (C) Abundances of linear alkanes generated under hydrothermal conditions. Lines are linear regressions of the C₇ to ∼C₂₁ alkanes, in which the chain growth probability (α) was 0.13. (D) Carbon isotopic composition of CH₄, C₂H₆, and C₃H₈.

As shown in Fig. 2A, the products were dominated by a homologous series of linear alkanes (SI Appendix, Fig. S4) with carbon chains of up to 24 carbons. Relative to linear alkanes, small amounts of alkenes (SI Appendix, Figs. S5–S7) and branched alkanes (SI Appendix, Figs. S8–S11) were also detected (Fig. 2B). Alkenes are important feedstocks to generate macromolecules or chiral compounds by secondary reactions. More importantly, the linear alkanes exhibited a regular log-linear decrease in abundance with an increasing number of carbons in the alkyl chain, commonly referred to as an Anderson–Schulz–Flory distribution (Fig. 2C). The ethane and propane display a regular pattern of decreasing ¹³C content with increasing carbon number but are all enriched in ¹³C relative to methane (Fig. 2D), which is consistent with the results starting from CO/HCO₂⁻ using the FTS mechanism (36, 37). Thus, the structures and relative abundances of long-chain hydrocarbons generated closely resembled those of long-chain carbons produced by a typical FTS. The selectivity of C₂ to ∼C₂₄ reached 24%. Considering that the primary Co minerals of wairauite (CoFe) and CoFe₂ are generated during serpentinization in the Earth’s ultramafic crust or upper mantle, the highest average cobalt content occurs in ultramafic rocks reaches a substantial ∼110 ppm (14–18), and also considering the geochemically feasible environment in this study, the results demonstrated herein provide direct evidence for the possible abiotic formation of hydrocarbons in natural alkaline hydrothermal systems.

Morphological Evolution of Co.

To further probe Co stabilization, the solid phase was analyzed by X-ray diffraction (XRD), as shown in SI Appendix, Fig. S12. The Co maintains its native valence state, while Fe is oxidized into magnetite (Fe₃O₄). Element-mapping (Fig. 3G) distribution also confirmed the existence of native Co and Fe oxide. Although Co maintains an unchanged valence state, interestingly, the morphology of Co changed completely. Initially, Co powder was used in the experiment (Fig. 3A), yet honeycomb Co nanosheets formed after a short reaction time of 10 to ∼30 min (Fig. 3 B and C) and were clearly observed for a longer reaction time of 1 to ∼3 h (Fig. 3 D and E). The change from the initial Co powder to the honeycomb nanosheets indicates that the native Co underwent oxidation (dissolution) and reduction (reprecipitation), resulting in no overall change in the valence state before and after the reaction, despite the change in morphology. As discussed before, although the reduction of CoO_x is difficult in industrial FTS, the hydrothermal reduction of CoO into Co is relatively readily reached through the assistance of HCO₃⁻, thus stabilizing Co through redox equilibrium.

Fig. 3.

Morphological evolution of Co in a hydrothermal environment. (A−E) Scanning electron microscopy (SEM) images of Co powder as a function of reaction time in the presence of Fe. Experimental conditions are the following: Fe, 80 mmol; NaHCO₃, 80 mmol; Co, 40 mmol; water filling, 50%; 300 °C, pH 8.23, 30 MPa. (F) SEM images of Co powder after reaction for 3 h without the addition of Fe. (Scale bars of A−F, 1 μm.) (G) Element mapping of the solid sample after reaction for 3 h.

Moreover, the H₂ produced in situ from Fe oxidation may serve as a template to assemble the dissolved/exfoliated Co into honeycomb nanosheets. A control experiment was performed, in which ex situ H₂ instead of Fe was used. As expected, no honeycomb Co nanosheets were observed (Fig. 3F). Thus, the in situ–produced H₂ from Fe oxidation achieved multiple purposes: reducing NaHCO₃ into long-chain hydrocarbons, reducing CoO into Co to maintain the stability of Co, and acting as a template to assemble the dissolved/exfoliated Co into a honeycomb structure. Moreover, the number of long-chain hydrocarbons (C₆ to ∼C₂₁) exhibited a sharp increase in the first 3 h (SI Appendix, Fig. S13), which was consistent with the formation of honeycomb nanosheets. These results clearly indicate that the formation of nanostructured Co accelerates C−C coupling. Also, the formation of long-chain hydrocarbons and nanosheets are not dependent on the initial Co morphology (SI Appendix, Fig. S14) and can be efficiently catalyzed by recycled Co−Fe₃O₄ (SI Appendix, Fig. S15).

We also experimented with different transition metals and metal oxides as catalysts at the same hydrothermal conditions, such as Ni, Cu, Mo, TiO₂, ZnO, and Al₂O₃, as additives (SI Appendix, Table S2) for abiotic synthesis because they may have occurred in the prebiotic alkaline hydrothermal fluids and constitute the main components of mineral deposits. The results show that only Co in the presence of Fe has the function to catalyze C−C coupling for long-chain hydrocarbon synthesis, demonstrating the irreplaceability of native Co as a catalyst that is also found in certain hydrogenations known from microbiological and chemical studies (39–41).

Mechanism of C−C Coupling to Long-Chain Hydrocarbons.

We next attempted to elucidate the mechanism of long-chain hydrocarbon formation in our hydrothermal system. One important question was whether our synthesis of long-chain hydrocarbons followed the traditional mechanism of FTS (i.e., is NaHCO₃ first reduced to CO which only then undergoes further reduction to a hydrocarbon?). Our gas chromatography–mass spectrometry (GC-MS) analysis of the gas sample did show that CO was formed (SI Appendix, Fig. S16), and furthermore, operando-attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy was developed (SI Appendix, Fig. S17) to confirm CO formation and the study of the mechanism. During heating, two peaks located at 2,128 and 1,935 cm⁻¹, clearly represented linear-type CO adsorbed on Co and Co−Fe interfacial sites, respectively (42–44). The gradual intensification of the peaks with increasing reaction temperature (Fig. 4A) also indicates the formation of CO. The experiment with CO as the initial carbon source for substituting NaHCO₃ under the same conditions also generated long-chain hydrocarbons (SI Appendix, Fig. S18). Thus, our hydrothermal NaHCO₃ reduction does follow a typical FTS mechanism via CO intermediate pathway.

Fig. 4.

Mechanism of NaHCO₃ reduction. (A) Operando ATR-FTIR spectra of NaHCO₃ reduction by Fe and Co metals. (B and C) XRD patterns of solid products (B) and the corresponding product distribution of HCOONa and long-chain hydrocarbons (C) as a function of initial Co amount. Experimental conditions are the following: Fe, 80 mmol; NaHCO₃, 80 mmol; water filling, 50%; 300 °C, 3 h, pH 8.23, 30 MPa. The vertical black lines in B are the XRD peaks of standard FeCO₃ (JCPDS No. 29–0696), Co (JCPDS No. 01–1278 for hcp and No. 15–0806 for fcc), and Fe₃O₄ (JCPDS No. 79–0418). (D) Proposed, competing pathways for the hydrothermal reduction of NaHCO₃ into HCOO⁻ (Left) and long-chain hydrocarbons (Right). In the absence of Co, only HCOO⁻ was formed by the reductive decomposition of FeCO₃. The addition of Co induced the transfer of the COOH intermediate accompanied by the generation of Fe−OH group, suppressing the formation of FeCO₃. The Fe−OH group enhanced the adsorption of CO, promoting the synthesis of long-chain hydrocarbons. JCPDS: Joint Committee on Powder Diffraction Standards.

However, the initial CO generated by NaHCO₃ reduction often reacted with OH⁻ to produce HCOO⁻ under alkaline hydrothermal conditions. Our current results and previous works also demonstrated that only HCOO⁻ was formed when Fe alone was used. It is known that enhanced CO adsorption on Co is key for C−C coupling in FTS; thus, our results indicate that CO can be stabilized on the Co surface under alkaline hydrothermal conditions with the use of Fe.

We further probed how the CO was stabilized by operando ATR-FTIR. Two significant peaks at 3,676 and 3,226 cm⁻¹, attributed to the surface hydroxyl groups on Fe (Fe−OH) (42, 45, 46), appeared and intensified gradually with increasing reaction temperature (Fig. 4A). Surface hydroxyl groups have been proposed to interact with CO₂/CO to enhance the CO₂/CO adsorption strength (47, 48). The lower wavenumber of the CO_ads band (1,935 cm⁻¹) indicates the stronger adsorption of CO on the Co surface (49), which may be attributed to the role of the Fe−OH group, but to further confirm the role of Fe−OH, experiments replacing Fe with H₂ in the presence of Co were performed. In these cases, the formation of long-chain hydrocarbons was not observed, indicating that Co alone has no catalytic activity. Therefore, Fe or its oxidation/hydroxylation products are a crucial part of the reduction mechanism herein. Together with the detection of Fe−OH, this result suggests that the oxidation products of Fe should involve Fe−OH entities such as green rust (50). Moreover, the peak from the hydroxyl group in Fe−OH almost disappeared in the absence of Co (SI Appendix, Fig. S19), indicating that the formation of Fe−OH was promoted by Co.

Experiments were further conducted by changing the initial amount of Co to study its influence on the Fe−OH formation. The formation of siderite (FeCO₃) was observed in the absence of Co (Fig. 4B); however, the XRD peak intensity of FeCO₃ decreased with increasing amounts of Co, and the peak of FeCO₃ completely disappeared when the amount of Co reached 40 mmol. Moreover, the distribution of the products between hydrocarbons and HCOO⁻ changed in response to the amount of Co catalyst (Fig. 4C). The formation of HCOO⁻ gradually decreased with increasing amounts of Co, and the synchronized trend of FeCO₃ and HCOO⁻, combined with the results of previous intensive studies (23), suggests that FeCO₃ is responsible for HCOO⁻ formation. The formation of FeCO₃ and its reductive decomposition into HCOONa and Fe₃O₄ was thermodynamically favorable [Δ_rG^θ (573 K) = −102.2 kJ ⋅ mol⁻¹].

In contrast, the number of long-chain hydrocarbons continued to increase with the Co amount (Fig. 4C). Therefore, the presence of Co reduced the formation of HCOO⁻ and promoted the formation of hydrocarbons by suppressing the formation of FeCO₃ and promoting Fe oxidation via Fe−OH. The reason that Fe oxidation prefers to form FeCO₃ rather than proceed via the Fe−OH pathway in the absence of Co is because the formation of Fe(OH)_x is thermodynamically less favorable [Δ_rG^θ (573 K) > 12.9 kJ ⋅ mol⁻¹ (x = 2)]. In the presence of Co, the formation of Fe−OH is attributed to the higher-adsorption energy of COOH on Co (−2.28 eV) (51) than that on FeO_x (−1.04 eV) (52), which promotes the transformation of the intermediate COOH reduced from FeCO₃ to the Co surface, leading to the formation of Fe(OH)_x.

Proposed Pathway of C−C Coupling versus HCOO⁻.

Based on the mechanism of C−C coupling, we propose a possible pathway of NaHCO₃ reduction to hydrocarbons using Fe and Co (Fig. 4D). First, HCO₃⁻ adsorbs on the surface of Fe because the formation of FeCO₃ is thermodynamically favorable, and the peak of the vibration of bound HCO₃ located at 1,340 cm⁻¹ was also observed both in the absence and presence of Co (SI Appendix, Fig. S19 and Fig. 4A). Second, the adsorbed HCO₃ is reduced to intermediate COOH, which then transfers to the Co surface, leading to the formation of Fe−OH groups, as evidenced by the cooccurrences of the Fe−OH peak at 3,676 cm⁻¹ and the COOH peak at 1,390 cm⁻¹. The peak at 1,390 cm⁻¹ can be assigned to the symmetric stretching mode of bound COOH on the Co surface (53), and this peak was not observed in the absence of Co. As mentioned before, the COOH intermediate prefers to adsorb on Co because of the higher-adsorption energy on Co (−2.28 eV) (51) than on FeO_x (−1.04 eV) (52), which suppresses the formation of FeCO₃. To confirm our assumption, we replaced Co with Ni and Cu, for which the adsorption energies of COOH are −2.25 and −1.52 eV (54), respectively. As expected, the formation of FeCO₃ was inhibited by increasing the adsorption energy of COOH on the metal surface (SI Appendix, Fig. S20), which further supported the transfer of COOH. Third, the bound COOH was reduced to CO, whose adsorption on Co is enhanced by Fe−OH to suppress the formation of HCOO⁻. Finally, the adsorbed CO undergoes sequential reduction and further polymerizes into long-chain hydrocarbons.

Notably, based on the proposed pathway of C−C coupling, the HCO₃⁻ is not directly involved in the abiotic synthesis. The HCO₃⁻ first reacts with Fe to form FeCO₃, which either undergoes reductive decomposition into HCOO⁻ or releases COOH to the Co surface. The proposed process is similar to the pathway that mantle-derived, inorganic carbon source takes rather than the one that seawater bicarbonate does, just as the former pathway has been shown to be directly responsible for the abiotic synthesis in an actual hydrothermal environment (2, 55, 56). Moreover, our system has both similar hydrothermal conditions and reaction pathways as those occurring in an actual hydrothermal environment, supporting the abiogenic origin of some hydrothermal petroleum.

Conclusions

In summary, our results show that hydrothermal reduction of NaHCO₃ can produce long-chain hydrocarbons up to C₂₄H₅₀ using general Fe and Co metals at ∼300 °C and 30 MPa. Bicarbonate is not only used as the carbon resource but also has a role of stabilizing Co, in which the honeycomb Co nanosheets are generated in situ. Operando ATR-FTIR revealed that the presence of Co promoted the formation of Fe−OH, which in turn enhanced the adsorption of the siderite-derived CO intermediate on Co for an efficient C−C coupling process. Taken together, we provide experimental evidence and insights into the mechanism of formation of long-chain hydrocarbons using native cobalt in the presence of Fe−OH in geologically feasible conditions on the early Earth. These findings may support the abiogenic hypothesis for some petroleum deposits, introduce an approach for nonnoble metal-catalyzed CO₂ conversion to petroleum fuel, and perhaps play into a better understanding of the emergence of life.

Materials and Methods

Chemicals.

Fe powder (325 mesh, ≥98%, Alfa Aesar Chemical Reagent), Co powder (≥99.9%, Sinopharm Chemical Reagent Co., Ltd; ≥99.9%, Aladdin Chemical Reagent; ≥99.8%, Alfa Chemical Reagent; ≥99.9%, Macklin Chemical Reagent), NaHCO₃ (AR, ≥99%, Sinopharm Chemical Reagent Co., Ltd), and H₂ (>99.995%, Shanghai Poly-Gas Technology Co.) were used.

Synthesis of Hydrocarbons.

The synthesis of hydrocarbons was performed using a batch-type microautoclave (∼42 cm³) system, lined with SUS 316 stainless steel, and equipped with a high-pressure valve for collecting gas. In a typical run, Fe powder (80 mmol), NaHCO₃ (80 mmol), ultrapure water (20 mL, water-filling level of 50%), and Co powder (40 mmol) were added to the reaction chamber. The reactor was sealed and treated by heating the autoclave to 300 °C at a heating rate of 20 °C/min with constant shaking in an induced heating furnace. After the treatment, the autoclave was removed from the furnace and cooled with an air blower to room temperature. Water filling was defined as the ratio of the volume of the water put into the reactor to the inner volume of the reactor.

Detection of Organic Compounds.

To detect the volatile hydrocarbons, the gas products were collected over a saturated NaCl solution and quantified by gas chromatography (GC) equipped with a thermal conductivity detector (Agilent GC7890A). To extract nonvolatile hydrocarbons, 5 mL dichloromethane (DCM) was injected into the reaction chamber after the reaction, and the reactor was shaken for 15 min. The DCM extract was analyzed by GC equipped with a mass spectrometry detector (Agilent GC7890A-MS5975C) for identification and a flame ionization detector (Agilent GC7890A) for quantification using an HP-5 column. The GC temperature program started at 35 °C for 5 min, ramped to 210 °C at 5 °C ⋅ min⁻¹, and was held at this temperature for 10 min. CO was confirmed by GC-MS with a molecular sieve column. The temperature program started at 35 °C for 2 min, ramped to 180 °C at 10 °C ⋅ min⁻¹, continued to ramp to 230 °C at 20 °C ⋅ min⁻¹, and was held at this temperature for 80 min. The organic acid anions in water were analyzed by GC-MS for identification and high-performance liquid chromatography (HPLC, Agilent 1260) for quantification. The GC-MS temperature program started at 40 °C for 4 min, ramped to 230 °C at 7 °C ⋅ min⁻¹, and was held at this temperature for 20 min. An Agilent 19091N-233HP-INNOWax Polyethylene Glyco column was used for sample separation. Helium (>99.999%) was used as the carrier gas. The HPLC was equipped with a tunable ultraviolet-visible detector adjusted at 210 nm and two Shodex RSpak KC-G columns. HClO₄ solution (2 mmol/L) was used as mobile phase with a flow rate of 0.8 mL/min.

Isotope Analysis.

Carbon isotope (δ¹³C) and concentration analyses of CH₄, C₂H₆, and C₃H₈ were conducted on a gas chromatograph coupled with isotope ratio mass spectrometer (IRMS, DeltaplusXP, Thermo Fisher Scientific) via a combustion furnace and an open split interface (GC Combustion III, Thermo Fisher Scientific (57). Briefly, the experiment was carried out at the following conditions: injection temperature 250 °C; split ratio 5:1; flow rate 1.5 mL/min; oven temperature from 50 (maintained 5 min) to 150 °C (maintained 10 min) at a rate of 10 °C/min. Pure helium (>99.99995%) was used as the carrier gas by passing through an HP-PLOT-Q column (30 m × 0.32 mm internal diameter, 10-µm film thickness). After leaving the GC column, the separated gas subsequently entered a combustion furnace (operating at 960 °C) containing a ceramic tube packed with CuO, NiO, and Pt wires and transformed into CO₂, which was then analyzed by the IRMS. The carbon isotope ratios were calibrated by the simultaneous injections of a laboratory-working standard gas, which beforehand has been calibrated by the international standard. The precision of our δ¹³C analysis is better than 0.5‰.

Catalyst Characterization.

After DCM extraction, the reaction precipitate was washed with deionized water and dried in a vacuum oven at 60 °C for further characterization. XRD patterns were obtained on an SHIMADZU XRD-6100 PC with Cu Kα radiation (λ = 1.54184 Å) in steps of 0.02° and an accumulation time of 0.6 s per step at 40 kV and 40 mA. The peak identifications were carried out based on the reference patterns reported in the Power Diffraction File published by the International Centre for Diffraction Data. Scanning electron microscopy photographs were taken on a Hitachi s-4800 electron microscope at an acceleration voltage of 5 kV.

Operando ATR-FTIR.

Operando ATR-FTIR was developed and performed to detect the intermediate in the NaHCO₃ reduction reaction. The spectra were recorded with a Nicolet FTIR spectrometer iS10 equipped with a liquid nitrogen–cooled, narrow-band mercury–cadmium–telluride detector and a PC-50–300-DFIR high-pressure cell. The experiments were conducted at an elevated temperature from 25 to 250 °C. Spectra were collected as difference spectra, with the initial state as the background spectrum.

Data Availability

All study data are included in the article and/or SI Appendix.