Geoelectrochemical CO production: Implications for the autotrophic origin of life

Efficient CO production realizing Wächtershäuser’s scenario was demonstrated in simulated early ocean hydrotherm systems.

Abstract

Wächtershäuser’s proposal of the autotrophic origin of life theory and subsequent laboratory demonstrations of relevant organic reactions have opened a new gate for the exploration of the origin of life. However, this scenario remains controversial because, at present, it requires a high pressure of CO as a source of carbon and reducing energy, although CO must have been a trace C species on the Hadean Earth. We show that, simulating a geoelectrochemical environment in deep-sea hydrothermal fields, CO production with up to ~40% Faraday efficiency was attainable on CdS in CO₂-saturated NaCl solution at ≤–1 V (versus the standard hydrogen electrode). The threshold potential is readily generated in the H₂-rich, high-temperature, and alkaline hydrothermal vents that were probably widespread on the early komatiitic and basaltic ocean crust. Thus, Wächtershäuser’s scenario starting from CO₂ was likely to be realized in the Hadean ocean hydrothermal systems.

INTRODUCTION

Since the first publication of Wächtershäuser’s theory of the autotrophic origin of life from primordial carbon fixation within a sulfide-rich hydrothermal vent (1, 2), a number of laboratory simulations for prebiotic organic synthesis have been performed within this scenario. Representative examples were published in Science more than a decade ago by Huber and Wächtershäuser: Under mildly hot, anaerobic, and alkaline aqueous conditions in the presence of metal sulfides (for example, FeS and NiS), they demonstrated the C–C bond formation that reminds us of the reductive acetyl–coenzyme A pathway (3), the reductive polymerization of HCN to α-hydroxy and α-amino acids (4), and the polymerization of α-amino acids to short peptides (5). Nevertheless, the significance of the autotrophic scenario has been questioned (6), because the proposed sulfide-pulled CO₂ reduction has never been realized: Huber-Wächtershäuser’s organic syntheses required a high pressure of CO (1 bar or more) as a source of carbon and reducing energy. The CO concentrations in the atmosphere and volcanic emissions on the Hadean Earth were likely to be trace (7) unless there was a huge supply of meteorite-derived reducing power (8) that would have inevitably destroyed the surface environment. Although Wächtershäuser maintained a life’s origin in a CO-rich volcanic environment in a later addition to his theory (9), the setting is likely incompatible with the conditions crucial to the organic synthesis (3–5) because volcanic environments are inherently acidic to neutral (10). Russell and colleagues (11–13) and, more recently, Lane and Martin (14, 15) have proposed an alternative theory that the primordial carbon assimilation took place in the Hadean ocean alkaline hydrothermal systems through the direct coupling of seawater CO₂ and hydrothermal H₂ on metal sulfides, but no clear experimental support has yet been provided (16, 17).

A different mechanism of abiotic CO₂ reduction in the early ocean hydrothermal systems has been suggested (18–20), for which redox, proton, and thermal gradients between the hydrothermal fluids and seawater across the sulfide deposits provide a continuous supply of reductive energy for inorganic carbon fixation. Using a sulfide chimney block obtained from a black smoker vent in the southern Lau Basin, Nakamura et al. (18) observed a long-distance electric conduction across the rock [0.2 to 9.1 ohms cm^–1 at 300 K; (21)] and its catalytic function for redox reactions at the mineral-seawater interface. Yamamoto et al. (19) verified the electricity generation in the deep-sea hydrothermal environment using a fuel cell coupling the oxidation of hydrothermal H₂S and the reduction of seawater O₂ in an artificial hydrothermal vent. In addition, the in situ electrochemical surveys of the Okinawa Trough hydrothermal fields (20) gave multiple lines of evidence for natural and widespread electricity generation via sulfide deposits. These laboratory and field investigations strongly suggest that geoelectrochemical energy and reactive surfaces are ubiquitous in the present-day hydrothermal systems and were even more abundant and widespread across the early ocean floor.

Here, we present a laboratory demonstration of efficient electroreduction of CO₂ on metal sulfides and discuss its relevance to the autotrophic scenario of the origin of life. We prepared low-crystalline sulfides of various metals (Ag, Cd, Co, Cu, Fe, Mn, Ni, Pb, or Zn) by mixing an alkaline solution of sodium sulfide and an acidic solution of the corresponding metal chloride. This mixing-induced sulfide formation likely occurred in the early alkaline hydrothermal vent environments and could comprise the main body of hydrothermal mineral deposits as the geoelectrochemical reactors (13, 22). Notice that the following experiments considerably augment recent studies by our group (23) and others (24), revealing the catalytic capability of some crystalline sulfides (for example, greigite). Here, we also studied molybdenum and tungsten sulfides (MoS₂ and WS₂) because of their importance in microbial metabolic reactions (25–27).

RESULTS

The CO₂ electrolysis was conducted in an H-type electrochemical cell that had two compartments separated by a proton exchange membrane, with a Ag/AgCl reference electrode and a carbon working electrode placed at one side, and a Pt counter electrode at the other side (fig. S1). The metal sulfide–deposited carbon electrode was exposed to a constant potential for 24 hours in 100 mM NaCl saturated with 1 atm CO₂ at room temperature (~25°C). Liquid-phase products were characterized by high-performance liquid chromatography (HPLC) and proton nuclear magnetic resonance (¹H NMR) spectroscopy, and gas-phase products were characterized by gas chromatography (GC). Figure 1 shows the Faraday efficiencies (FEs) of CO₂ reduction to CO (CO₂ + 2H⁺ + 2e^– → CO + H₂O) and to formate (CO₂ + H⁺ + 2e^– → HCOO^–) at –0.8, –1.0, and –1.2 V [versus the standard hydrogen electrode (SHE)]. CO and formate were the sole carbon products observed with the FE > 0.1%, along with H₂ that evolved from the competing proton reduction (2H⁺ + 2e^– → H₂). The CO formation occurred effectively on CdS and Ag₂S at –1.0 and –1.2 V (versus SHE), whereas negligible amounts of CO were produced in other experimental conditions or in the absence of sulfide. Additional experiments with varied potential (Fig. 1C) revealed that CO was generated at an electric potential below –0.8 V (versus SHE) on both sulfides. The FE shifted positively with decreasing potential and attained around 40% on CdS and 30% on Ag₂S at ≤–1.0 and ≤–1.1 V (versus SHE), respectively. For formate, active formation was observed only on CuS (FE ~4% at –1.0 and –1.2 V versus SHE; Fig. 1B). These product characteristics were largely different from those on the corresponding pure metals (28), indicating the significance of sulfur to the surface-CO₂-electron interactions.

Fig. 1. Electrochemical CO₂ reduction on various metal sulfides in 0.1 M NaCl saturated with 1 atm CO₂ at room temperature (~25°C).

(A and B) FEs for CO₂ reduction to (A) CO (CO₂ + 2H⁺ + 2e^– → CO + H₂O) and (B) HCOO^– (CO₂ + 2H⁺ + 2e^– → HCOO^–) at –0.8, –1.0, and –1.2 V versus the SHE. (C) CO production efficiencies on CdS and Ag₂S as a function of electric potential. Error bars correspond to the overall reproducibility of the experimental data that was evaluated from three independent experiments.

To examine whether the electrochemically produced CO serves as a driving force for the subsequent abiotic organic synthesis, we collected the reductive gas generated on CdS at –1.0 V (versus SHE) that was primarily composed of CO (ca. 40%) and H₂ (ca. 60%) (Fig. 1C), sealed it in a serum bottle containing a solution of simple carbon compounds (methanethiol, potassium cyanide, or glycine) and NiS, and incubated it in a similar manner as was reported by Huber and Wächtershäuser (3–5) (see the Supplementary Materials). Our experiments largely reproduced the original reports, although there were smaller yields with fewer varieties of products owing to the lower initial CO concentration (fig. S2). We did not carry out a “one-pot” organic synthesis in an electrochemical cell because dissolved neutral species inevitably penetrate into the counter electrode side and undergo oxidation at uncontrollable potential, which makes data interpretation difficult. The occurrence and combination of multiple reactions are likely to be more realistic than single-pot reaction in natural hydrothermal vent environments. The complexity of the geochemical system could also allow the accumulation of organic compounds (29) that are generated under a continuous supply of CO by the sulfide-catalyzed CO₂ electroreduction.

DISCUSSION

Were the experimental conditions examined above geologically and geochemically feasible in the Hadean ocean hydrothermal systems? Our experiments revealed that CO production on metal sulfides occurs at electric potentials below –0.8 V (versus SHE). In submarine hydrothermal vents, electrons are catalytically provided by the oxidation of reductive chemicals at the fluid-mineral interface (20). If H₂S is assumed to be the sole electron source in the fluid, the potential is always higher than the threshold (–0.8 V versus SHE) in a wide range of temperature and pH (Fig. 2, A and B). However, if H₂ is present at a concentration of 1 mmol kg^–1 or higher (Fig. 2, C and D), H₂ oxidation in hot and alkaline pH conditions can provide the potential well below the desirable value [for example, –0.98 V (versus SHE) at 200°C and pH 12 in the presence of 1 mmol kg^–1 H₂]. The H₂-enriched hydrothermal fluid systems have been hypothesized to occur abundantly on the Hadean and early Archean ocean floor; two representative systems are a komatiite-hosted high-temperature type at oceanic ridges and plateaus (Fig. 3A) (30) and a low-temperature type hosted by komatiite or peridotite at ridge flank regions (13). A recent hydrothermal experiment of komatiite in CO₂-saturated NaCl solution demonstrated that high concentrations of CO₂ suppress H₂ generation during low-temperature serpentinization (<250°C) owing to the formation of Fe²⁺-bearing dolomite, whereas the high-temperature serpentinization (>300°C) can generate extraordinarily high concentrations of H₂ even with CO₂-saturated seawater (31). Under such high-temperature conditions, the dehydration of epidote to anorthite followed by the precipitation of calcite controls the pH and makes it highly alkaline (pH ≥12), which has been demonstrated by a hydrothermal alteration experiment of basalt in CO₂-rich seawater (32). It seems likely that alkaline hydrothermal systems were widespread across the early ocean crust and floor that were dominated by basalt and/or komatiite (33). Meanwhile, the early basalt-hosted hydrothermal systems with CO₂-rich seawater may have also generated a variety of hydrothermal fluids with different H₂ concentrations and pH values. With a total large surface area of basalt-seawater reaction in the hydrothermal circulation from recharge to discharge, the seawater CO₂ would be quickly removed, and the CO₂-depleted water/rock reaction could generate acidic to neutral, metal-rich (and occasionally H₂-rich) black smoker fluids as observed today (Fig. 3B) (32).

Fig. 2. Thermodynamically predicted electric potentials of fluids from deep-sea hydrothermal systems.

Redox potentials of (A) H₂S (1 mmol kg^–1), (B) H₂S (10 mmol kg^–1), (C) H₂ (1 mmol kg^–1), and (D) H₂ (10 mmol kg^–1) as a function of pH at 25°, 100°, 200°, and 300°C calculated assuming the half reactions of S + 2H⁺ (or H⁺) + 2e^– → H₂S (or HS^–) [for (A) and (B)] and 2H⁺ + 2e^– → H₂ [for (C) and (D)]. S (solid sulfur) was chosen as the H₂S (or HS^–) oxidation product because the H₂S/S redox couple provides a major potential control on the sulfide-rich hydrothermal vent environments (19). See the Supplementary Materials for the calculation procedure.

Fig. 3. Abiotic carbon fixation in the primitive hydrothermal system.

(A) The hot and active mantle convection in the primitive Earth induced a lot of massive upwelling of komatiite/basaltic melts beneath the ocean floor (30), (B) where interactions between the solidified (ultra)mafic rocks and the CO₂-rich seawater with different water/rock ratios led to a variety of end-member fluid chemistry, including the H₂-rich, high-temperature (T), and alkaline type and the metal-rich, high-temperature, and acidic (or neutral) ones (31, 32). (C) On the ocean floor, mixing of the hydrothermal fluids and seawater generated sulfide-rich chimneys (22), and the potential gradient across the chimney drove a continuous electron flow. The electric potential at the chimney-seawater interface could reach less than –1 V (versus SHE) in alkaline hydrothermal vent environments. The low potential, in the presence of sulfides rich in Cd²⁺ and Ag⁺, allowed the electrochemical CO₂ reduction to CO with the FE as high as dozens of percent, together with H₂ evolution. The produced CO served as a driving force for the subsequent abiotic organic synthesis that preceded the origin of life as schematically indicated in the figure.

Metals from the present-day black smoker vents have been observed to emanate over thousands-of-kilometers scale across hydrothermal plumes under soluble and/or colloidal forms with a limited vertical diffusion (34). Wider and denser metal distributions in the early deep ocean are envisioned (35) because of the anaerobic, Fe-rich, and slightly acidic character of ancient seawater (36, 37) with much greater global thermal convection than the present level (38). Cd²⁺ is a trace but ubiquitous hydrothermal dissolved metal ion (39) and has a very poor solubility in alkaline sulfide conditions (40). Therefore, Cd²⁺ and CdS dispersed from the early basalt-hosted hydrothermal systems would have precipitated and incorporated in the sulfide deposits in the closely localized alkaline hydrothermal vent environments.

There is no definitive constraint on the dissolved sulfide flux from the early ocean alkaline hydrothermal vents. If the fluid sulfide concentrations were within or lower than the micromolal level, the fluid-seawater mixing is expected to precipitate Cd, Zn, and Cu selectively against Ni, Co, and Mn owing to their difference in sulfide solubility (40). FeS precipitation would not have occurred even when a high seawater Fe²⁺ level (50 μmol kg^–1) had been attained (40). In the present-day and Paleozoic hydrothermal sulfide deposits, Cd is observed primarily as an isomorphous impurity in sphalerite (ZnS), rather than a single CdS crystal, with the Cd/Zn substitution ratio of up to 13.2 weight % (41, 42). The Cd/Zn ratio tends to increase at higher precipitation temperature and lower concentrations of dissolved sulfide and chloride (43, 44); hence, hot and alkaline hydrothermal systems could favor the formation of Cd-enriched ZnS. Naturally generated multiple metal-sulfide precipitates in some cases exhibit higher catalytic efficiencies for the electroreduction of CO₂ than the individual sulfides (23). Thus, some of the hydrothermal sulfide deposits of the H₂-rich, high-temperature, and alkaline hydrothermal systems that were probably widespread in the early komatiitic and basaltic ocean floor (Fig. 3, A and B) are considered to provide favorable conditions for electrochemical CO production. This environmental situation is fully consistent with that assumed in Wächtershäuser’s abiotic organic synthesis (Fig. 3C) (3–5). Therefore, this study provides the first experimental evidence that the presence of the key player in Wächtershäuser’s scenario for the autotrophic origin of life, CO, was possible under a geologically and geochemically plausible setting on the primitive Earth.

The geoelectrochemical CO production demonstrated in this study could also enhance the theory for the origin of membrane bioenergetics (14). A number of microorganisms, in both archaea and bacteria, are known to use CO and formate as energy and carbon sources (45, 46). With the redox potential of a CO/CO₂ pair (E⁰ = –520 mV), which is even lower than that for H₂/H⁺ (E⁰ = –414 mV), CO can solely reduce ferredoxin (E⁰ = –400 mV) and generates a larger quantity of adenosine triphosphate (ATP) in the course of the autotrophic carbon assimilation than the H₂/CO₂ pair does (46). The energetic advantage of the CO metabolism would have brought an extra benefit for the origin and early evolution of chemolithotrophic systems in the early ocean hydrothermal vent environments.

At least two issues remain to be addressed: Why only Ag and Cd sulfides can effectively reduce CO₂ to CO, and whether methanethiol and cyanide were really available. Ag₂S and CdS are characterized by a relatively high energy conduction band edge (47) and by a relatively weak metal-S bond strength (48) among the sulfides examined in this study. The former property likely facilitates electron transfer from the charged surface to the adsorbed oxidants (H₂O and CO₂) (49). The latter energetically favors the C–S bond formation through reconnecting the metal-S bond, facilitating the CO₂ conversion to CO by stabilization of the reaction intermediate (50). Although these advantages qualitatively explain the high CO productivities of CdS and Ag₂S (Fig. 1), a definitive understanding of our results requires detailed structural and electronic study of adsorbed species. The availability of methanethiol and cyanide anion remains controversial (51, 52), although several mechanisms have been proposed for their abiotic formation in hydrothermal systems (53, 54). Because electrochemistry is a frequently used process to synthesize organic compounds with C–N and C–S bonds (55), the geoelectrochemical settings could have favored the production of reactive C1 compounds from naturally ubiquitous carbon, nitrogen, and sulfur sources (for example, CO₂, N₂O, and SO₂) under naturally feasible conditions.

MATERIALS AND METHODS

Preparation of metal sulfides

All metal sulfides except for WS₂ and MoS₂ were prepared by a dropwise addition of 100 mM Na₂S_(aq) [or 50 mM Na₂S_(aq) for the case of Ag₂S] into a 100 mM aqueous solution of the corresponding metal chloride (AgCl, CdCl₂, CoCl₂, CuCl₂, FeCl₂, MnCl₂, NiCl₂, PbCl₂, or ZnCl₂) under vigorous stirring with a final volume ratio of 1:1. The valence states of metals are all stable under the estimated Hadean ocean redox condition [0.432 V (versus SHE)] (56). Solid precipitates were then separated from the supernatant by centrifugation and were dried under vacuum. To prevent oxidation by atmospheric O₂, the sample preparation was conducted in a globe box filled with N₂ gas (>99.9999%), with 4% H₂ being added (the COY system). In many cases, the synthesized sulfides showed broad x-ray diffraction signals, with low intensity indicating low crystallinity (fig. S3). All chemicals were purchased from Wako as reagent grade. Deaerated Milli-Q water (18.2 megohms) was used as the solvent. WS₂ and MoS₂ were obtained from Sigma-Aldrich (99%, ~90 nm) and were used without further purification.

Electrochemical experiments

Figure S1 schematically illustrates the electrochemical cell. For each experiment, about 50 mg of metal sulfide catalyst was deposited on the carbon working electrode (5.7 cm²), deaerated 100 mM NaCl_(aq) was poured into the cell (60 ml to the working electrode side and 10 ml to the opposite side), and pure CO₂ gas (>99.995%) was allowed to flow (20 ml min^–1) into the solution for at least 1 hour before each experiment. While keeping the CO₂ flowing, a constant potential was applied on the carbon electrode by using a multi-potentiostat (PS-08; Toho Technical Research). CO₂ buffered the solution slightly acidic (pH = 5.75 ± 0.25) during the electrolysis. All potentials were measured against an Ag/AgCl reference electrode in saturated KCl and converted to the SHE scale using the following equation

$$\mathit{E} (\text{vs}. \text{SHE})=\mathit{E} (\text{vs}. \text{Ag}/\text{AgCl})+0.198\mathrm{V}$$ (1)

After 24 hours of duration, the sample solution was filtered with a polytetrafluoroethylene (PTFE) membrane filter (pore size, 0.22 μm) and was analyzed by HPLC and ¹H NMR spectroscopy. The gas-phase product was collected in a polypropylene gas trap immersed in the solution and was analyzed by GC. The gas sampling was carried out for a certain fixed period (note that the sampling period was varied depending on the activity of the reductive gas generation), for which bubbles generated on the metal sulfides were accumulated into the gas trap while intrusion of the flowed CO₂ was carefully avoided. The gas products collected after various durations of the electrochemical experiment showed mutually similar compositional characteristics over the first 24 hours (within the overall reproducibility; ±10%). The FE was calculated from the following equations

$$\text{FE} =\frac{\text{charge required for the production}}{\text{total charge built up for the first} 24 \text{hours}}$$ (2)

for the liquid-phase products, and

$$\text{FE} =\frac{\text{charge required for the production}}{\text{total charge built up during the sample collection}}$$ (3)

for the gas-phase products.

Organic synthesis experiments

Three types of organic synthesis experiments were conducted according to the methods reported by Huber and Wächtershäuser (3–5), with the electrochemically generated reductive gas on CdS at –1 V (versus SHE) replacing pure CO gas.

(i) NiSO₄ ⋅ 6H₂O (131 mg, 0.5 mmol) was deaerated in a serum bottle (100 ml) sealed with a butyl rubber cap and an aluminum stopper, and 2.5 ml of aqueous solution containing Na₂S (39 mg, 0.5 mmol), CH₃SNa (1.8 mg, 25 μmol), and NaOH was added for the in situ precipitation of NiS. The bottle was then filled with the reductive gas (80 ml, 1 atm) and was heated at 100°C for 7 days. The sample suspension was then filtered with a PTFE membrane filter and analyzed by liquid chromatography–mass spectrometry (LC-MS) for identification of organic acid products and by HPLC for their quantification. The filtrate had a pH of 8.6, which was measured by a portable pH meter (Seven2Go Pro, Mettler Toledo).

(ii) A serum bottle (100 ml) was charged with Ca(OH)₂ (0.5 g, 6.7 mmol), closed, deaerated, and then charged with 2.5 ml of aqueous solution containing NiSO₄ (77 mg, 0.5 mmol) and KCN (33 mg, 0.5 mmol) together with the reductive gas (80 ml, 1 atm). After heating at 100°C for 5 days, the sample suspension was filtered and analyzed for organic acid and amino acid products. The filtrate showed a pH of 12.6.

(iii) A serum bottle (100 ml) was charged with NiSO₄ ⋅ 6H₂O (131 mg, 0.5 mmol), closed, deaerated, and then charged with 2.5 ml of aqueous solution containing Na₂S (78 mg, 1 mmol) and glycine (9.4 mg, 125 μmol) together with the reductive gas (80 ml, 1 atm). After heating at 100°C for 1 day, the sample suspension was filtered and analyzed for glycine peptides. The filtrate showed a pH of 7.9.

Sample analysis

Organic acid products were quantified by using a Shimadzu HPLC system equipped with an electric conductivity detector and an anion exchange column (Shim-pack SCR-102H, Shimadzu) set at 40°C. The p-toluenesulfonic acid solution (5 mM) was used as the eluent at a rate of 1.6 ml min^–1.

¹H NMR spectra were acquired using a Bruker Avance III spectrometer (400 MHz) at the sample temperature of 303.0 K. Typically, 0.45 ml of sample solutions was mixed with 0.05 ml of D₂O (99.9%; Merck Millipore) containing 5 mM 3-(trimethylsilyl)-1-propanesulfonic acid-d₆ sodium (DSS-d₆; Sigma-Aldrich) and was placed in an NMR tube (5 mm outside diameter; Wilmad-LabGlass). DSS-d₆ was used for the calibration of the 0-ppm (parts per million) position and to quantify the product concentrations as an internal standard. A solvent suppression was run to minimize the solvent signal. The obtained spectra of sample solutions after the CO₂ electrolysis (fig. S4) showed an intense signal of HCOO^– at 8.45 ppm when CuS was used as a catalyst, whereas no significant peak except for H₂O and DSS-d₆ appeared on the other spectral regions and in the other sample spectra. The quantified concentrations of HCOO^– coincided well with the corresponding HPLC results (within ±10%); therefore, the FE was calculated using the HPLC data and was presented in Fig. 1B.

The GC measurements were conducted using a GC-4000 (GL Sciences) equipped with a HayeSep Q column (GL Sciences) and a Shincarbon-ST column (Shinwa). He (>99.99995%) was used as the carrier gas at a flow rate of 24 ml min^–1. The column temperature was initially kept at 40°C for 6 min, raised to 200°C at a rate of 20°C min^–1, kept for 6 min, raised to 250°C at a rate of 50°C min^–1, and then kept for 20 min. H₂, CO, CH₄, CO₂, C₂H₆, and C₃H₈ were observed with the retention times of ca. 2.1, 5.0, 8.8, 11.8, 16.7, and 28.0 min, respectively. A pulsed discharge detector was used to quantify the concentrations of CO, CH₄, C₂H₆, and C₃H₈, and a thermal conductivity detector for the H₂ concentration. Some gas samples were also analyzed by a gas chromatograph mass spectrometer (GCMS-QP2020; Shimazu) equipped with MICROPACKED-ST column (Shinwa). Results obtained by the different systems were consistent with each other within the error (±10%).

Amino acid and peptide products were analyzed using a Jasco HPLC system equipped with post-column derivatization with o-phthalaldehyde and a fluorescence detector operated at 345 nm for excitation and at 455 nm for emission. Five citrate buffer solutions of different citrate concentrations and pH values were used as eluents in a stepwise condition. A cation-exchange column (AApak Na II-S2, Jasco) was used at 50°C.

LC-MS measurements were conducted with a Xevo Qtof MS platform (Waters) combined with a UPLC system (Acquity, Waters). To enhance the ionization of small acids while suppressing the interference with inorganic salts, positive-mode electrospray ionization combined with reversed-phase LC was performed after derivatization with 2-nitrophenylhydrazine (57). In the LC separation using a CORTECS UPLC C₁₈ column (Waters), 0.1% trifluoroacetate and acetonitrile were used as the eluents for binary gradient elution with the flow rate at 0.3 ml min⁻¹ and the column temperature at 35°C. The acetonitrile volumetric ratio was initially held at 10%, increased linearly to 40% and then to 90% for 2 to 6 min and 6 to 6.7 min, respectively, and held at 15% for 7 min. The time-of-flight analyzer was set to a sensitivity mode with a resolving power of 20,000 at the mass/charge ratio (m/z) of 556 corresponding to protonated leucine encephalin, and the m/z range of 20 to 1500 was calibrated with sodium formate.

Thermodynamic calculation

The redox potentials of H₂S and H₂ as a function of temperature and pH (Fig. 2) were calculated, respectively, assuming the following half reactions

$$2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2$$ (4)

$$\mathrm{S}+2{\mathrm{H}}^{+}(\text{or} {\mathrm{H}}^{+})+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2\mathrm{S} (\text{or} {\text{HS}}^{-})$$ (5)

using the equations

$${\mathit{E}}_{\mathrm{h}}=\frac{1}{2\mathit{F}}(2\mathit{R}\mathit{T}\text{ln}{\mathrm{\alpha }}_{{\mathrm{H}}^{+}}-\mathit{R}\mathit{T}{\mathrm{\alpha }}_{{\mathrm{H}}_2}-{\mathrm{\Delta }}_{\mathrm{f}}{\mathit{G}}^{\mathrm{o}}({\mathrm{H}}_2))$$ (6)

$${\mathit{E}}_{\mathrm{h}}=\frac{1}{2\mathit{F}}((1+\mathit{x})\text{RTln}{\mathrm{\alpha }}_{{\mathrm{H}}^{+}}-\text{xRTln}{\mathrm{\alpha }}_{{\mathrm{H}}_2\mathrm{S}}-(1-\mathit{x})\mathit{R}\mathit{T}\text{ln}{\mathrm{\alpha }}_{{\text{HS}}^{-}}+{\mathrm{\Delta }}_{\mathrm{f}}{\mathit{G}}^{\mathrm{o}}(\mathrm{S})-\mathit{x}{\mathrm{\Delta }}_{\mathrm{f}}{\mathit{G}}^{\mathrm{o}}({\mathrm{H}}_2\mathrm{S})-(1-\mathit{x}){\mathrm{\Delta }}_{\mathrm{f}}{\mathit{G}}^{\mathrm{o}}({\text{HS}}^{-}))$$ (7)

In these equations, T, R, and F stand for temperature in kelvin, the gas constant (8.31447 J mol^–1 K^–1), and the Faraday constant (96485 J mol^–1 V^–1), respectively. α_i represents the activity of the species i, and x signifies the mole fraction of H₂S ($=\frac{{\mathrm{M}}_{{\mathrm{H}}_2\mathrm{S}}}{{\mathrm{M}}_{{\mathrm{H}}_2\mathrm{S}}+{\mathrm{M}}_{{\text{HS}}^{-}}}$; ${\mathrm{M}}_{{\mathrm{H}}_2\mathrm{S}}$ and ${\mathrm{M}}_{\mathrm{H}{\mathrm{S}}^{-}}$ denote the molality of H₂S and HS^–, respectively). In addition, Δ_fG^o(i) represents the standard partial molal Gibbs free energy of formation of the species i at desired temperature and pressure, which were calculated according to the revised Helgeson-Kirkham-Flowers (HKF) equations of state (58) together with the thermodynamic data and the revised HKF parameters reported by Shock et al. for H₂ and H₂S (59) and HS^– (60). The temperature and pressure dependences of Δ_fG^o for S (solid sulfur) were calculated as follows

$$\mathrm{\Delta }{\mathit{G}}_{\mathit{P},\mathit{T}}^{\mathrm{o}}=\mathrm{\Delta }{\mathit{G}}_{{\mathit{P}}_{\mathrm{r}},{\mathit{T}}_{\mathrm{r}}}^{\mathrm{o}}-{\mathrm{S}}_{{\mathit{P}}_{\mathrm{r}},{\mathit{T}}_{\mathrm{r}}}^{\mathrm{o}}(\mathit{T}-{\mathit{T}}_{\mathrm{r}})+{\displaystyle {\int }_{{\mathit{T}}_{\mathrm{r}}}^{\mathit{T}}}{\mathit{C}}_{{\mathit{P}}_{\mathrm{r}}}^{\mathrm{o}}\mathrm{d}\mathit{T}-\mathit{T}{\displaystyle {\int }_{{\mathit{T}}_{\mathrm{r}}}^{\mathit{T}}{\mathit{C}}_{{\mathit{P}}_{\mathrm{r}}}^{\mathrm{o}}} \text{dln}\mathit{T}+{\displaystyle {\int }_{{\mathit{P}}_{\mathrm{r}}}^{\mathit{P}}{\mathrm{V}}_{\mathit{T}}^{\mathrm{o}}}\mathrm{d}\mathit{P}$$ (8)

where $\mathrm{\Delta }{\mathit{G}}_{{\mathit{P}}_{\mathrm{r}},{\mathit{T}}_{\mathrm{r}}}^{\mathrm{o}}$ and ${\mathrm{S}}_{{\mathit{P}}_{\mathrm{r}},{\mathit{T}}_{\mathrm{r}}}^{\mathrm{o}}$, respectively, represent the standard molal Gibbs energy and entropy at the reference temperature (T_r = 298.15 K) and pressure (P_r = 1 bar). ${\mathit{C}}_{{\mathit{P}}_{\mathrm{r}}}^{\mathrm{o}}$ represents the standard molal heat capacity at P_r, and ${\mathit{V}}_{\mathit{T}}^{\mathrm{o}}$ denotes the standard molal volume at the temperature of interest. In the present calculation, the values of $\mathrm{\Delta }{\mathit{G}}_{{\mathit{P}}_{\mathrm{r}},{\mathit{T}}_{\mathrm{r}}}^{\mathrm{o}}$, ${\mathit{S}}_{{\mathit{P}}_{\mathrm{r}},{\mathit{T}}_{\mathrm{r}}}^{\mathrm{o}}$, and ${\mathit{C}}_{{\mathit{P}}_{\mathrm{r}}}^{\mathrm{o}}$ as a function of temperature for S were taken from Robie et al. (61), whereas the value of ${\mathit{V}}_{\mathit{T}}^{\mathrm{o}}$ was assumed to be constant in the range of temperature and pressure of our interest. The value of x is expressed as

$$\mathit{x}=\frac{{\mathrm{\gamma }}_{{\text{HS}}^{-}}{\mathrm{\alpha }}_{{\mathrm{H}}^{+}}}{{\mathrm{\gamma }}_{{\text{HS}}^{-}}{\mathrm{\alpha }}_{{\mathrm{H}}^{+}}+{\mathit{K}}_{{\mathrm{H}}_{\mathrm{s}}\mathrm{S}}{\mathrm{\gamma }}_{{\mathrm{H}}_{\mathrm{s}}\mathrm{S}}}$$ (9)

where γ_i represents the activity coefficient of the species i (α_i = M_i × γ_i) and ${\mathit{K}}_{{\mathrm{H}}_{\mathrm{s}}\mathrm{S}}$ is the dissociation constant of H₂S (H₂S → HS^– + H⁺), whose value was calculated as

$${\mathit{K}}_{{\mathrm{H}}_{\mathrm{s}}\mathrm{S}}= \text{exp}\left(\frac{{\mathrm{\Delta }}_{\mathrm{f}}{\mathit{G}}^{\mathrm{o}}({\mathrm{H}}_2\mathrm{S})-{\mathrm{\Delta }}_{\mathrm{f}}{\mathit{G}}^{\mathrm{o}}({\text{HS}}^{-})}{\mathit{R}\mathit{T}}\right)$$ (10)

In all calculations, the values of γ_i were calculated by using the extended Debye-Hückel equation (62) with the ionic strength of 0.1 (NaCl). The pressure was set to 100 bar. S^2– and the ion pair NaHS were not considered in this calculation because these are expected to be minor S and/or Na species in the examined aqueous conditions (63, 64).

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/4/eaao7265/DC1

fig. S1. A schematic of the electrochemical cell.

fig. S2. Abiotic organic synthesis driven by the electrochemically generated reductive gas on CdS at –1.0 V (versus SHE) in 100 mM NaCl saturated with 1 atm of CO₂.

fig. S3. X-ray diffraction patterns of metal sulfides.

fig. S4. ¹H NMR spectra of the CO₂-saturated 0.1 M NaCl after applying –1.2 V (versus SHE) for 24 hours in the presence of metal sulfides.

table S1. Summary of total current densities and FEs for CO₂ reduction on metal sulfides.

table S2. Peak list of x-ray diffraction patterns of metal sulfides.