Research ArticleGEOCHEMISTRY

Photoreduction of inorganic carbon(+IV) by elemental sulfur: Implications for prebiotic synthesis in terrestrial hot springs

See allHide authors and affiliations

Science Advances  18 Nov 2020:
Vol. 6, no. 47, eabc3687
DOI: 10.1126/sciadv.abc3687


Terrestrial hydrothermal systems have been proposed as alternative birthplaces for early life but lacked reasonable scenarios for the supply of biomolecules. Here, we show that elemental sulfur (S0), as the dominant mineral in terrestrial hot springs, can reduce carbon dioxide (CO2) into formic acid (HCOOH) under ultraviolet (UV) light below 280 nm. The semiconducting S0 is indicated to have a direct bandgap of 4.4 eV. The UV-excited S0 produces photoelectrons with a highly negative potential of −2.34 V (versus NHE, pH 7), which could reduce CO2 after accepting electrons from electron donors such as reducing sulfur species. Simultaneously, UV light breaks sulfur bonds, benefiting the adsorption of charged carbonates onto S0 and assisting their photoreduction. Assuming that terrestrial hot springs covered 1% of primitive Earth’s surface, S0 at 10 μM could have produced maximal 109 kg/year HCOOH within 10-cm-thick photic zones, underlying its remarkable contributions to the accumulation of prebiotic biomolecules.


The origin of life remains unsolved but can be summarized as organic monomers formed, accumulated, assembled, developed into complex polymers and, finally, living beings (14). A critical step is the prebiotic synthesis of simple organic molecules represented by formic acid, acetic acid, pyruvate, and amino acids from inorganic carbon sources. This process must have sustainably supplied base materials for further inoculation of complex life substances, which ultimately developed into primitive life forms (2, 5, 6). The pathways and environments that could have given rise to primitive organic molecules, therefore, are of great importance and have been the subject of many models and hypotheses.

One hypothesis of interest is that the terrestrial origin of life and nascent life emerged in vapor-dominated zones of inland geothermal systems (710). Unlike submarine hydrothermal vents, the geothermal fluids that well up from terrestrial hot springs develop on land or in shallow pools, accompanied with volcanic landmasses, geysers, and steam vents. With the discovery of hyperthermophiles as well as thermal, pH, and chemical redox gradients (1012), terrestrial hydrothermal systems are also regarded as a likely environment for abiogenesis and even the cradle of early life. However, the initial biological records in terrestrial systems are not well preserved because of long-term geological and anthropogenic activities. The earliest signs of life on land (~3.5 billion years ago) discovered in Western Australia (13) could be regarded as an evidence for early life’s possible origin in terrestrial hydrothermal environment. Nevertheless, until now, there has been no convincing experimental model for the creation of necessary organic molecules from terrestrial inorganics.

Notably, early aqueous environments including the pools and springs developing in terrestrial hydrothermal systems are presumed to involve considerable carbonates [H2CO3, HCO3, and CO32−, denoted as carbon(+IV)], which were formed by the dissolution of atmospheric CO2 that was present at high concentration (14, 15). Whether organics could be synthesized from these inorganic carbon species is an important issue and worth further exploration, wherein minerals have been suggested to play key roles (3, 6, 1619). In particular, semiconducting sulfide minerals present in hydrothermal systems, like sphalerite (ZnS) and alabandite (MnS), were demonstrated to be able to catalyze the reduction of CO2 and the synthesis of primitive biomolecules under ultraviolet (UV) light (20, 21). These semiconducting metal sulfides have thus been suggested as the key to unravel the puzzle of life’s origin in terrestrial hot springs (7). However, in the primitive Earth, a majority of sulfide minerals were deposited at deep-sea subduction zones and ridges (22). A few of them with low solubility precipitated at terrestrial hydrothermal systems, while those that precipitated slowly like ZnS and MnS were suggested to be enriched in far-off ponds and puddles away from hot springs, fed by downstream and cooled fluids (7). Therefore, it is very probable that the origin of life in early terrestrial hydrothermal systems is dominated by another unrevealed mineral model.

In modern volcanic hydrothermal systems on the surface of Earth, elemental sulfur (S0) is a quite abundant mineral (2326). It was also the main product of photochemical or comproportionation reactions that involved −2 and +4 valent sulfur in the primitive Earth (2729). The discharged hydrogen sulfide (H2S) and sulfur dioxide (SO2) gas from prebiotic craters and hydrothermal vents were substantial and estimated to be over three times as abundant as those from modern volcanos (28, 30, 31). The production rate of S0 aerosol in a proposed model reached about 107 molecules cm−2 s−1; some of these aerosols could have condensed into particles and then dropped into the early ocean (30). Therefore, a scenario in which natural S0 was immersed in the carbonate environment, especially in terrestrial hydrothermal systems, was likely to exist in the prebiotic Earth. However, S0 has rarely been an attractive catalyst or reactant but was rather deemed a negligible by-product in atmospheric or aquatic photochemistry (27, 30). As a typical nonmetallic semiconducting mineral (32), whether S0 could photocatalyze some prebiotic synthesis reactions and thus contribute to the chemical origin of life has not yet been reported and elucidated.

In this study, we provide experimental evidence for the photocatalytic production of HCOOH in a simplified terrestrial hydrothermal system with dissolved carbonate and suspended S0. The semiconducting and photoelectrochemical properties of S0 are systematically investigated, and the results support the model that sufficient photoelectrons with highly reducing potentials are excited to enable CO2 reduction under UV light. In addition, UV light is shown to break the sulfur bonds of S0, which favors the adsorption of carbonate anions onto S0 and thus facilitates charge transfer. This study highlights the remarkable role of semiconducting S0 in converting carbon(+IV) into organics and supports a possible pathway for biomolecule accumulation in terrestrial hydrothermal systems. The distinguished highly reducing photoelectrons from UV-excited S0 could be further extended to synthesize more complicated organics, thus playing comparable roles to the traditional highly reducing agent like H2 and high temperature–dependent pathway in hydrothermal systems.


Geochemistry of Tengchong terrestrial hot spring with abundant S0

Tengchong geothermal field, located in Yunnan Province in southwestern China, is famous for its Cenozoic volcanism and present-day geothermal features. In these active hot springs and pools (58° to 97°C) exposed on the surface of Earth, massive S0 with characteristic yellow color was observed to deposit along the margin of the hot pool below the water’s surface (Fig. 1, A and B). Mineral phase identification by x-ray diffraction revealed orthorhombic α-sulfur (homocyclic S8, which is the most stable form of S0) as the main component in the submerged yellow deposit, accompanied with griceite (LiF) and amorphous geyserite (Fig. 1C). Some white deposits, however, lacking S0 and consisting of thenardite (Na2SO4), halite (NaCl), griceite, and geyserite, were deposited beyond the spring water (Fig. 1,B and C). At some fumaroles without charging water, a small amount of S0 was deposited on the surface of porous siliceous sinter (Fig. 1D). Under scanning electron microscopy (SEM), the S0 mineral in hot springs had a crystal size of several micrometers that did not show fixed crystal shape (Fig. 1, E and F). In comparison, the S0 at fumaroles was present in bigger and well-crystallized druses (Fig. 1D). Multiple ions, mainly including Na+, K+, Mg2+, Li+, Cl, S2−, and SO42−, were detected in all six sampling sites (table S1). The ascending H2S gas could maintain a reducing environment for the preservation of S0 in the water. The white deposits, which were dominated by sulfates and were not immersed in reducing water, therefore, were the oxidation products of the yellow deposits. As can be seen from above observations, one of the most remarkable features in the terrestrial hydrothermal system is the deposition and suspension of S0 in the sunlit and spring water.

Fig. 1 The occurrence of S0 at a geothermal field in Tengchong, China.

(A) A degassing spring named Dagunguo (96°C, pH 8.0) that is 5 m in diameter with S0 deposition labeled by red arrows. (B) S0 at the poolside and in the hot water. (C) X-ray diffraction patterns of yellow and white deposits collected from (B). a.u., arbitrary unit. (D) Active fumarole with abundant S0, siliceous sinter, and salt deposition (the digital thermometer shows 73.4°C). Inset: Well-crystallized S0 druse observed by SEM. (E) SEM image of S0 particles collected in hot springs. (F) Sulfur element mapping pattern at S Kα line corresponding to the region in (E). Photo credit: Y.W., Peking University.

Photoreduction of carbonate to produce HCOOH in the presence of S0

The photochemical reactions were conducted in three control systems: system 1 with S0 and UV light, system 2 with S0 and visible light, and system 3 with UV light only. As the reaction progressed, HCOOH was continuously produced only in system 1 (Fig. 2A and fig. S1), while no HCOOH was detected in either system 2 or system 3. The amounts of HCOOH increased along with the concentration of carbon(+IV) and S0 (Fig. 2B). After irradiation for 12 hours, 2.6 mM HCOOH was produced in the solution containing 50 mM NaHCO3 and S0 (5 g/liter).

Fig. 2 The concentration of HCOOH varied with the following conditions.

(A) Irradiation time in three control systems. (B) Different initial concentration of NaHCO3 and S0. (C) Different initial pH. (D) Different light intensity (254-nm UV light). Carbonate species ratios at corresponding pH values are illustrated in (C). IPCE under 254-nm irradiation based on HCOOH production is plotted in (D). Unless otherwise specified, the uniform reaction system contained S0 (5 g/liter) and 50 mM NaHCO3 at pH 6 and was irradiated for 12 hours in a wide UV-band mercury lamp.

In the presence of both S0 and UV light, the total yields of HCOOH after 12 hours were 0.6, 2.6, 0.8, and 0.2 mM at pH 4, 6, 8, and 10, respectively (Fig. 2C). Notably, the chemical forms of carbonate species drastically varied with pH. The H2CO3 or HCO3 dominated systems (99 and 97% in pH 4 and 8, respectively) produced moderate HCOOH, while the system with 61% CO32− and 39% HCO3 (pH 10) produced the minimum. The highest yields of HCOOH, unexpectedly, occurred in a weakly acidic system (pH 6) with 64% H2CO3 and 36% HCO3, suggesting that both H+ and HCO3 were crucial to carbon(+IV) reduction in this case.

Controlled experiments were conducted under three monochromatic wavelengths: 254, 313, and 365 nm, corresponding to UVC (200 to 280 nm), UVB (280 to 320 nm), and UVA (320 to 400 nm), respectively. Notably, HCOOH was only produced under 254-nm irradiation, and its yields were positively correlated with light intensity (Fig. 2D). The incident photon-to-current conversion efficiency (IPCE) under 254-nm irradiation was calculated to be 7.6, 7.0, 6.1, and 4.5‰, corresponding to a light intensity of 0.25, 0.50, 0.75, and 1.00 mW/cm2, respectively (Fig. 2D). All the above results suggested that the production of HCOOH required the involvement of both S0 and UVC light, and its yielding rate was correlated with the concentrations of C(+IV) and S0 (Fig. 2B), pH (Fig. 2C), and light intensity (Fig. 2D).

The photoactivity of S0 under UV light

Production of photoelectrons. The optical absorption spectra of S0 showed different patterns at 25° and −200°C (Fig. 3A). At room temperature, a clear absorption edge emerged in the visible band (Fig. 3A) and its onset was 475 nm (λ1). When the sample was cooled down to −200°C, the absorption edge moved to 280 nm (λ2), while the previous edge at 475 nm disappeared (Fig. 3A). It was reported that the color of S0 changes from yellow to white at low temperature (33). The absorption spectra confirmed this behavior and suggested that the optical absorbance of S0 is temperature dependent. In terms of its emission spectra, dual fluorescence signals were observed to be situated almost at the center of two absorption edges (Fig. 3A). Therefore, S0 should have two types of electronic transition, denoted as transition I and transition II, which occur under visible and UV light, respectively. The band structure calculation based on density functional theory (DFT) proves that S0 (in the form of orthorhombic S8) is an indirect semiconductor with a narrow indirect bandgap and a wide direct bandgap (fig. S2). Transition I in Fig. 3A is assigned to the indirect mode, which absorbs parts of visible light (λ < 475 nm) or equivalent energy of photons higher than 2.6 eV. Transition II only responds to UV light (λ < 280 nm) with higher energy (4.4 eV) and is the direct mode. The indirect transition is phonon dependent and thus needs high enough temperature to generate phonons, leading to its inefficiency at low temperature (Fig. 3A).

Fig. 3 Optical characteristics and bonding change of S0 under UV light.

(A) Absorption spectra (in black) at 25°C and −200°C, and room temperature emission spectrum (in blue) ranging from UV to visible band. The dashed lines in the emission spectrum curve were subtracted from the multiple-frequency peaks. (B) Recording of current changes under alternating on and off states of a xenon lamp [with UV and visible light (VL)] and the lamp with a UV filter, using S0 or a blank substrate as electrodes. (C) Raman spectra recorded under 325- and 532-nm light irradiation, respectively. (D) Time-course EPR and the simulated spectra of S0 under UV light.

Notably, the photo-generated electrons and holes are produced in the conduction band (CB) and valence band (VB), respectively. To confirm the formation of photo-generated carriers (electrons and holes) under light, photoelectrochemical measurements were performed on S0 to compare the photocurrent change with and without light irradiation (Fig. 3B). Under xenon lamp (mainly visible light with 4% UV component), the current produced by S0 (net current between the on and off state of the light) was 1.5-fold higher than the background current produced by the blank substrate. When the UV light was filtered, however, the photocurrents showed no notable difference between S0 and the substrate, indicating that S0 could not be excited without UV irradiation. Therefore, the effective electron transition only occurred in UV-excited direct mode rather than in the indirect mode, which can be attributed to the phonon-independent process of the former mode. According to the UV photoelectron spectra (fig. S3 and table S2), the VB and CB edges of S0 were determined to be −6.86 and −2.46 eV [absolute vacuum energy scale (AVS)], respectively. This result is highly consistent with the recent DFT calculation (34). Accordingly, the redox potentials of CB electrons and VB holes were converted into −2.34 and 2.06 V [versus normal hydrogen electrode (NHE)] at pH 7, respectively.

Breaking of S─S bonds. The Raman spectrum of S0 with the feature of homocyclic S8 at 532 nm had stretching vibration modes at 438 and 473 cm−1 and bending modes at 154, 182, 219, and 246 cm−1 (Fig. 3C). When excited at 325 nm, however, a completely different spectrum that was assigned to the chain-like polymeric sulfur (denoted as Sμ) was recorded. Compared with the Raman bands of S8, the frequencies of two stretching vibrations of Sμ decreased by 20 cm−1, while those related to bending modes increased to 260 and 272 cm−1. This was attributed to the longer bond length (2 pm) and decreased bond angle (2°) of Sμ than S8 (35).

To further investigate the change of chemical bonds in S0 under UV light, an in situ UV-irradiated electron paramagnetic resonance (EPR) measurement at −170°C was conducted. Some recorded signals (Fig. 3D), which gradually increased with UV exposure time, consisted of three components with anisotropic features. According to the g tensor values (g1 = 2.041, g2 = 2.025, and g3 = 2.003), the simulated spectrum given in Fig. 3D was identical to the experimental one, indicating that the three signals were derived from a single paramagnetic radical. Notably, the radical was previously detected in a similar EPR measurement and reported to be the end of sulfur chain with a trans conformation (36). This confirmed that the chemical bonds between two sulfur atoms were broken under UV light and thus the homocyclic rings in S0 were converted into chain-like S0 allotropes (Sμ is one of the possible configurations). Notably, the EPR signals faded immediately when the light was turned off (Fig. 3D), which implied that the two short-lived dangling bonds would probably rebind and return to the initial state. Therefore, the fracture of the S─S bond, just like the excitation of photoelectrons, could only be activated by UV light.

Adsorption of carbonate molecules and formation of formate on S0. In situ attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR) was used to monitor the molecules on the surface of S0 in UV-irradiated carbonate solution. It showed that some peaks emerged after a long period of UV irradiation (Fig. 4). These features are mainly assigned to the monodentate carbonates or bicarbonates at 1518 cm−1 (collectively denoted as HCO3(ad) due to HCO3 as the dominant species under this near-neutral condition), HCO3 (1419 and 1219 cm−1), and HCOOH (1395 cm−1) (37, 38). The indication of carbonate molecules bonded on S0 (i.e., HCO3(ad)), and the corresponding formation of HCOOH strongly confirmed the chemical adsorption of carbonate molecules on S0 during their UV-induced photoreduction. There was no evidence for bonding of carbonate molecules on the surface of S0 in the absence of UV light.

Fig. 4 ATR-FTIR spectra of molecules detected on the surface of S0 in UV irradiation.

The spectrum collected in the dark with no absorption feature remained unchanged for more than 30 min.


Reaction mechanisms based on the semiconducting properties of S0

On the basis of the temperature-dependent optical absorption spectra (Fig. 3A) and DFT calculation (fig. S2), S0 was characterized as an indirect semiconductor with two electronic transition modes, i.e., a phonon-independent direct one with a bandgap of 4.4 eV and a phonon-assisted indirect one with a bandgap of 2.6 eV. This is consistent with some previous spectroscopic studies (39, 40). The photocurrent from UV-irradiated S0 was high, while that produced under visible light was negligible (Fig. 3B), suggesting that the formation and transition of photoelectron-hole pairs in the UVC-induced direct transition were much more efficient than the indirect one excited by visible light, as described by Eq. 1.

The band structure of S0 is illustrated in Fig. 5A, in which the potentials of its VB, CB, and some redox couples are included for comparisons (see detailed values in tables S2 and S3). The reduction of CO2 to HCOOH is widely accepted to be a multielectron transfer process (41). The formation of an active carbon(+IV) intermediate (denoted as CO2∙−(ad)) after one single-electron transfer to the adsorbed CO2 (denoted as CO2(ad); Eq. 2) is recognized as the rate-limiting step with a very negative potential of −1.9 V (versus NHE). CO2∙−(ad) could further accept one electron and two protons to finally yield HCOOH (Eq. 3). Notably, the energy potential of photoelectrons from S0 is situated at −2.34 V (versus NHE at pH 7), much more negative than the potential for CO2 activation (Eq. 2). The reduction of CO2 is thus thermodynamically feasible by the photocatalysis of S0. The results in Fig. 2C show that the maximum yield of HCOOH occurs in a slightly acidic environment (pH 6), confirming to some extent the significance of protons. Particularly, the concentration of S0 oxidation products including sulfate (SO42−) and sulfite (SO32−) also increased with time in UV light (fig. S4), just as HCOOH behaved. Because the VB of S0 is situated at 2.06 V (versus NHE at pH 7), much more positive than SO32−/S0 (−0.03 V) and SO42−/SO32− couples (−1.36 V) (42) (Fig. 5A), the photo-generated holes would undergo a successive two-step oxidation reaction that first oxidizes S0 itself to SO32− (Eq. 4 and Fig. 5B) and further to SO42− (Eq. 5 and Fig. 5B). Because of the different rates of these two steps, S0 and SO32− constitute the competing hole scavengers, thus resulting in the lower concentration of SO32− than SO42− as shown in fig. S4. Reducing species like S0 and SO32−, if present in sufficient amounts in surrounding environments, thus not only serve as electron donors for CO2 reduction but also help separate photoelectron-hole pairs to release free photoelectrons.S0λ<280 nme+h+(photoelectron-hole pair production)(1)CO2(ad)+eCO2·(ad) (single-electron transfer to CO2(ad))(2)CO2·(ad)+e+2H+HCOOH (proton-assisted electron transfer to CO2·(ad))(3)S0+4h++3H2OSO32+6H+(holes scavenged by S0)(4)SO32+2h++H2OSO42+2H+(holes scavenged by SO32)(5)

Fig. 5 Schematic of photocatalytic mechanism.

(A) Band structure model of S0 showing the two transition modes, and the redox potentials of VB, CB, and some related redox couples (versus NHE, pH 7). e and h+ refer to photo-generated electrons and holes, respectively. Specific potential values are listed in tables S2 and S3. (B) Adsorption, activation, and photoreduction of carbonates by S0 under UV light. For simplification, S0 crystal is illustrated as a homocyclic S8 molecule.

When the incident photon energy is not high enough for a direct transition but sufficient to cause an indirect transition (i.e., 280 nm < λ < 475 nm), only a few photoelectrons would be produced in the indirect band. This happens because the electron transition mode is phonon-assisted and thus has a small absorption coefficient (fig. S2). In this situation, the reduction of CO2 is thermodynamically inhibited, and thus, no organic products can be produced.

Reaction mechanisms based on broken bonds reacting with adsorbed molecules

The S─S covalent bonds in S0 can be broken by UV light (Fig. 3, C and D), as described in Eq. 6. Simultaneously, carbonate species can be adsorbed onto the surface of UV-irradiated S0 (Fig. 4). All these observations suggest that the surface of S0 would become quite active and highly absorbent to polar molecules and charged ions (e.g., HCO3 and CO32−) once the nonpolar bonds are broken under UV light and become polar radicals, as shown in Fig. 5B and described by Eq. 7. The monodentate CO32− or HCO3 is thus the actual adsorbed carbon(+IV) species on the surface of S0. The recent DFT calculations reveal that the valence electrons of Sn species (n ≤ 8) after chain cleavage from S8 ring are dominantly localized at the terminal chain atoms (34). Such robust radicals on the surface of S0 would potentially play dual roles in the aqueous carbonate system. On the one hand, the unpaired electrons of the radicals can be directly trapped by the adsorbed groups and thus assist the reduction of targeted molecules. On the other hand, these radicals preferentially adsorb charged HCO3 and CO32−, and thus accelerate the indirect transfer of photoelectrons from S0 to carbon(+IV) (Fig. 5B). From this point of view, an alkaline (pH > 8) environment gives rise to HCO3 and CO32−, and thus might have been more beneficial to carbon(+IV) reduction than the acidic environment. However, the thermodynamic advantage from protons for carbon(+IV) reduction should be taken into consideration (Fig. 5A). In this case, a balance is achieved in that the weakly acidic carbonates containing H2CO3 and HCO3, e.g., 64% H2CO3 and 36% HCO3 at pH 6, could produce the maximal amount of HCOOH (Fig. 2C).SSUV lightS·+S·(cleavage of sulfur bonds)(6)S·+HCO3HCO3(ad) (adsorption of charged carbonates)(7)

Implications for photoreactive S0 in prebiotic terrestrial hydrothermal systems

Modern hydrothermal systems have been considered as a potential window to glance at the prebiotic Earth (13, 43, 44). We observed that solid S0 mineral occurred abundantly in modern terrestrial hydrothermal systems, such as Tengchong in China (Fig. 1), Yellowstone National Park in United States (23), and other sites in Japan (24), Indonesia (25), and New Zealand (26). As the main comproportionation reaction product of H2S, SO2, and soluble sulfur species (23, 45, 46), S0 minerals were abundant and widespread in the terrestrial hot springs and lakes in modern Earth. While drawing an analogy with modern cases and performing model simulations (2831), the concentration of S0 in early terrestrial hot springs was likely to be over µM level, by estimating from its sufficient sources including high-level atmospheric S0 aerosol and dissolving sulfuric anions at µM to mM level. Different from primitive metal sulfides that precipitate mainly at deep-sea subduction zones and ridges (22), S0 should be distributed primarily in the shallow photic zone of primitive hot springs due to its hydrophobicity and low density (23, 24). Notably, in the primitive Earth, extreme UVC with an eightfold greater total flux than nowadays could penetrate an ozone-free atmosphere to reach the surface of Earth (47, 48) and even reach the depth of several centimeters to meters in early ocean (30, 49). Elemental sulfur in terrestrial hydrothermal systems and its absorption of extreme UVC could have laid the foundation for the occurrence of photocatalysis in the primitive Earth.

HCOOH is the main detectable organic species in the UV-irradiated system (Eqs. 2 and 3), where S0 provided the reducing photoelectrons and was consumed by oxidation to sulfite and sulfate by a self-sacrifice catalytic process. As one of the most essential precursors in the abiotic synthesis of hydrocarbons (50, 51), HCOOH could be produced from carbon(+IV) reduction by H2 in analog submarine hydrothermal conditions (19, 51). Our work shows that solar light can provide the energy required for the abiogenic reduction of carbon(+IV) in terrestrial hydrothermal systems where the temperature is less than 100°C. In particular, semiconducting minerals like S0 are remarkable mediators by absorbing and converting solar energy into the chemical form. We can identify two advantages for this kind of energy-transforming pathway. On the one hand, the potential of photoelectrons from S0 (−2.34 V versus NHE at pH 7) when excited by UVC is much more negative than that of H2 (−0.42 V for H2O/ H2 couple at pH 7) and the photoelectrons from sulfide minerals like ZnS (−1.64 V) and MnS (−1.61 V) (52), providing S0 with extraordinary reducing power for HCOOH synthesis. On the other hand, there were substantial reducing ionic species like S0, SO32−, S2−, Fe2+, and NO2 in the primitive ocean, which could not directly reduce carbon(+IV) under ordinary conditions due to thermodynamic energy barriers. After donating electrons to scavenge the photo-generated holes of UVC-excited S0 (2.06 V versus NHE at pH 7), however, those electrons would be converted to highly reducing photoelectrons and thus surmount the energy barriers required for carbon(+IV) reduction. Based on this, it can be concluded that the weakly reducing species also contribute to carbon(+IV) reduction in systems that are exposed to sunlight and contain the mineral photocatalyst S0.

To estimate the abiotic production of organic molecules and evaluate the role of S0 proposed in this work, a simple model with S0 particles suspended in the photic zone of the early terrestrial hot springs is assumed. The concentration of HCOOH produced each day (denoted as CHCOOH, mM/day) is plotted against S0 concentration (denoted as ms, mM) in fig. S5, obtained in the experiment using simulated prebiotic conditions: 130 mM carbonate ions (15), pH 6.0 (15, 53), and 6 W·m−2 UV flux ranging from 200 to 300 nm (20, 48). After linear regression, the mathematical relationship between CHCOOH and ms is well described by linear Eq. 8. Furthermore, the estimation of the total yields (absolute quantity) of HCOOH is made by considering the thickness of the UVC photic zone (denoted as DUV) and the area ratio of hot springs to the total surface of early Earth (denoted as A). The product of these two parameters determined the total volume of UVC-irradiated S0 suspensions (more detailed information in Materials and Methods). Accordingly, the maximum mass of cumulative HCOOH depending on ms, A, and DUV is illustrated in Fig. 6, which is made based on Eq. 9. Notably, in a reasonable set of conditions represented by ms (up to 10 μM), A (less than 10%), and DUV (up to 1 m), the maximum amount of HCOOH produced in this model is comparable to other organics with similar molecular weight like formaldehyde, hydrogen cyanide, and amino acids, which have been proposed from other abiogenic sources, e.g., atmospheric photochemistry, impact shocks, and lightning, respectively (54). For instance, if assuming the primitive terrestrial hot springs account for 1% of Earth’s surface, S0 at 10 μM would produce maximal 1.71 × 109 kg of HCOOH per year within 10-cm-thick photic zone. Note that the conversion of HCOOH to other complicated biomolecules has a moderate reducing potential like −0.11 V (versus NHE at pH 7) for methane and 0.07 V for acetic acid (21, 41). The further synthesis of macromolecules based on HCOOH is thermodynamically favored under the photocatalysis of S0. Furthermore, S0 might also be involved in the so-called cyanosulfidic chemistry and contribute the yield of building blocks for life (55). Therefore, the extraordinarily reducing photoelectrons of S0 could provide an alternative and efficient abiogenic pathway to produce biomolecules in primitive hot springs.CHCOOH(mM/day)=0.02 mS(8)CHCOOH(kg/year)=1.71×1012DUV·mS·A(9)

Fig. 6 Cumulative yields of HCOOH in the proposed model as a function of thickness of photic zone, concentration of S0, and the area ratio of primitive terrestrial hot springs to Earth’s surface.

Some proposed main sources of abiogenic synthesis of organics are presented for comparison.

Terrestrial hydrothermal systems are distinguished by the everyday input of sufficient solar energy, multiple reducing ions as electron donors, and the abundant semiconducting minerals like S0. By virtue of these conditions, the lower-temperature terrestrial hydrothermal systems could also support primordial carbon fixation and energy conversion, which are comparable to the traditional H2-dependent pathways near high-temperature submarine hydrothermal vents. Moreover, the moderate temperatures in terrestrial hydrothermal systems and the absorption of extreme UV light by S0 could prevent biomolecules from decomposition, providing suitable habitats for the incubation of primitive life. Besides the well-known clays, metals, and metal sulfides, S0 thus turns out to be a newfound nonmetallic mineral catalyst that can enable the prebiotic synthesis and preservation of biomolecules. Beyond Earth, UV-activated S0 minerals may also catalyze the conversion of inorganic carbon into organic molecules on exoplanets. On Mars, for example, siliceous sinter deposits were discovered by the rover Spirit (56, 57), implying a terrestrial hydrothermal system in Martian surface and near-surface environments. The newly discovered photocatalysis of S0 would provide a novel clue for seeking or detecting biomolecules and biosignatures.


Preparation of S0 samples

Natural samples collected from the hot springs were dried at 40°C for 12 hours and then ground into powders. Natural S0 druses collected from the fumarole were selected in a stereoscope equipped with a 20× objective for roughly removing foreign materials. Commercial S0 powders (99.9 weight %) were purchased from Sinopharm Chemical Reagent Beijing Company Limited. Single-crystal S0 was from Geological Museum of Peking University, which was a natural sample but lacked exact information about its origin. Three types of S0 film on different substrates were prepared. The S0 film electrode on conductive glass (F-doped SnO2, abbreviated as FTO) was prepared by spin-coating method: Commercial S0 was first dissolved in carbon disulfide (CS2). Then, a piece of 2.5 cm by 2.5 cm transparent FTO glass was spun using a spin coater (KW-4A, China) with a rate of 4000 rpm. S0-saturated CS2 solution was dripped onto the spinning FTO glass for 40 s, on which a thin S0 film was left after the evaporation of CS2. The second S0 film was deposited on a single-crystal Ge that was already embedded into an ATR accessory. S0-saturated CS2 solution was dripped onto the Ge crystal, and the S0 film was obtained after naturally evaporating CS2. Another S0 film on quartz glass was prepared by thermal evaporate plating method: Commercial S0 powders were put into a tubular furnace and heated at 500°C. The produced hot sulfur vapor was flowed with the nitrogen to the unheated part, where a 2-mm-thick quartz wafer glass with 13 mm in diameter was put in advance. A thin S0 film was obtained from the solidified sulfur vapor. The specific uses of each S0 sample are shown in table S4.

Mineralogical identification and morphological observation of S0 samples

Natural samples were identified using an x-ray diffractometer (X’Pert Pro MPD, PANalytical, The Netherlands) equipped with Cu Kα irradiation (λ = 1.5406 Å). The patterns were recorded from 5° to 85° (2θ) with a scanning speed of 2°/min.

Micro-Raman analysis was carried out on a spectrometer (inVia Reflex, Renishaw, UK) equipped with a 532-nm Nd:YAG laser and a 325-nm He-Cd laser. The beam was focused within 1 μm onto samples at the microscope stage through a ×50 objective.

The morphology of natural S0 was observed by environmental SEM (Quanta 200FEG, FEI, USA) equipped with energy-dispersive X-ray spectroscopy detector at 15.0 kV. The colloidal solution in photochemical experiments was collected and dripped onto a holey carbon film supported by a Cu grid. After air-drying, the prepared sample was loaded into the holder of transmission electron microscope (Tecnai F20, FEI, USA) and observed at 200 kV. Digital micrograph version 3.6.5 (Gantan Ltd.) was applied for the image processing.

Optical and semiconducting characterization of S0

The absorbance of S0 powder sample at room temperature in 200 to 750 nm was detected on a spectrometer (UV-3600 Plus, Shimadzu, Japan) equipped with a diffuse integrating sphere attachment. BaSO4 was used as a reference, and the slit width of the incident light was 2 nm. The absorbance of S0 under low temperature was recorded in the transmission mode. S0 was cooled to −200°C by liquid nitrogen, which was controlled by an intelligent temperature controller (MercuryiTC, Oxford Instruments, UK). The room temperature fluorescence was collected with an F-7000 FL spectrophotometer (Hitachi, Japan). The 200-nm exciting light passed through a 5-nm slit and reached the S0 sample; the slit for the emitting light was 7 nm.

UV photoelectron spectroscopy of S0 on FTO substrate was carried out on an ESCALAB 250Xi spectrometer (Thermo Fisher Scientific, USA) with HeI (21.22 eV) emission line used as the excitation source. The data were collected with a helium pressure of about 2 × 10−8 mbar in the analysis chamber and at −10 V bias.

The determination of energy levels of VB and CB for S0

The work function (Φ) (Fermi level in vacuum) of S0 was calculated by Eq. 10, where hv equals 21.22, and EF and Ecutoff refer to the Fermi level and the kinetic energy of secondary electron cutoff, respectively. For calibration, the value of EF was referenced to the measured value of an Au electrode, which directly contacted S0 during the measurement. As shown in fig. S3, EF (from referenced Au) and Ecutoff were estimated as 20.78 and 3.76 eV, respectively. The work function of S0, based on Eq. 10, was estimated as 4.2 eV on the AVS. Therefore, its Fermi level was −4.20 eV (versus AVS), and the energy positions of VB edge (versus AVS) were calculated as −6.86 eV according to the gap between EF and VB (2.66 eV). The CB edge (versus AVS) of S0 was at −2.46 eV due to its direct bandgap of 4.4 eV.

Note that the energy levels of a semiconductor on the AVS correspond to those obtained in an electrolyte at the pH condition where its surface is zero charge (i.e., pHPZC) (52, 58). Therefore, the redox potentials of electrons in CB and holes in VB could be calculated based on Eqs. 11 and 12 (52), where ENHE and EAVS refer to potentials with respect to the NHE and energy levels on AVS, respectively, and EpH is the potential at any given pH value. According to the pHPZC value of S0 (ca. 2.0) (59), potentials of electrons and holes for S0 in aqueous electrolytes could be calculated as −2.34 and 2.06 V (versus NHE) at pH 7 (table S2).hvΦ=EFEcutoff(10)ENHE=EAVS4.50(11)EpH=EPZC+0.059×(pHPZCpH)(12)

Band structure calculation based on DFT

DFT-based first-principles calculation was performed using the Vienna Ab initio Simulation Package (VASP 5.4.4) (60) together with the ion-electron interaction described by the projector augmented wave method. Bandgap value was obtained with the HSE06 hybrid functional. A primitive cell of α-S with 32 atoms and six electrons (3s23p4) for valence electrons of each S atom were modeled. The Hartree-Fock mixing parameter and the cutoff energies of plane wave were set to the standard values of 25% and 340 eV, respectively. During VASP calculations, the gamma-centered k-point grid 4 × 4 × 4 was used for geometry optimization. The force convergence criterion was set as 0.01 eV Å−1 when the lattice parameters were relaxed, and the criterion for energy convergence when atomic positions were relaxed was set as 0.01 meV.

In situ experiments on S0 on UV irradiation

In situ X-band EPR measurement was conducted on a JES-FA200 spectrometer (JEOL, Japan). A fraction of single-crystal S0 was put into a quartz tube, and the sample chamber was cooled down to −170°C by liquid nitrogen. A high-pressure mercury lamp (USH-500SC, JEOL, Japan) was used to emit UV light. Each spectrum collection lasted 2 min, and time-course signals were recorded after the UV light was on.

In situ FTIR experiments were performed on a Vertex 70 spectrometer (Bruker, Germany) equipped with a horizontal Ge-based ATR accessory. Before the ATR-FTIR measurements, 1 ml of solution, which contained 50 mM NaHCO3, was adjusted to pH 6 by H3PO4, and then dripped onto the prepared S0 film on Ge crystal. A background spectrum was taken, and another spectrum was recorded using this background several minutes later to ensure that the system was in equilibrium. A high-pressure mercury lamp (CEL-M500, Aulight, China) was placed 10 cm over the ATR accessory. The UV light was on once the system was in equilibrium. During the irradiation, 64 scans at a resolution of 4 cm−1 in the range of 1000 to 4000 cm−1 were collected for each ATR-FTIR spectrum using OMNIC software.

Photocurrent measurements in an electrochemical system

Photocurrent measurements were carried out in a quartz glass reactor with three-electrode configuration using CHI 660C electrochemical apparatus (Chenhua, China). The prepared S0 electrode, a platinum sheet (1 cm by 1 cm), and Ag/AgCl (3 M KCl) electrode were set as the working electrode, counter electrode, and reference electrode, respectively. The mixture of 0.5 M Na2SO4 and 0.5 M methanol was used as the electrolyte. The applied bias was set at 1.2 V (versus Ag/AgCl). An external xenon lamp (PLS-SXE300, Perfectlight, China) with simulated solar light spectrum directly irradiated the side of the working electrode without covering any S0. The irradiation power density reaching the reactor was kept constant (6 mW/cm2) after adding a UV filter. Lights on and off were controlled by the shielding separator in the same time period (60 s).

Photochemical experiments on carbonate reduction

Batches of experiments were carried out in simplified terrestrial hydrothermal conditions, in which only commercial S0, sodium bicarbonate (NaHCO3), pH, and UV light were considered. The reaction temperature was kept at ~35°C when irradiated with UV light. All preparation and sampling operations were performed in a glove box (Universal, MIKROUNA, China) filled with high-purity argon [the concentration of O2 less than 0.1 parts per million (ppm)]. A quartz vessel reactor containing S0, 50 ml of oxygen-free water, and NaHCO3 was anaerobically sealed and fastened with an aluminum cap. Unless otherwise specified, the uniform conditions were set as S0 (5 g/liter), 50 mM NaHCO3 (pH 6), and irradiation for 12 hours in high-pressure mercury lamp (CEL-M500, Aulight, China). The pH was adjusted by 6 M H3PO4 and 1 M NaOH. The experimental and blank groups (without S0 or light) were performed in triplicate. All reagents were of analytical grade and prepared using oxygen-free water. The oxygen-free water was prepared as follows: the distilled deionized water (18 megohm·cm) in a glass bottle was boiled for 20 min and then tightly sealed when it was boiling; it was then moved to the glove box for replacing the residual air with argon (99.999%) for 24 hours before use. In particular, several kinds of light sources were used for investigating the role of light wavelength in reactions, including wideband UV light from a mercury (Hg) lamp, monochromatic UV light, wideband solar light from xenon lamp, and wideband visible light from a light-emitting diode lamp. The light intensity (table S5) and emission spectra (fig. S6) of all lamps were measured with a radiometer (FZ-A, Photoelectric Instrument Factory of Beijing Normal University, China) and a fiber spectrometer (S3000, Seemantech, China). The transmittance of quartz apparatus measured with a spectrometer (UV-3600 Plus, Shimadzu, Japan) is illustrated in fig. S7. More than 87% of UV light ranging from 250 to 380 nm, which was emitted from those lamps in experiments, would not be absorbed by the quartz apparatus.

Identification of products

The concentration of dissolved anions was determined by ion chromatography (ICS-1100, Thermo Fisher Scientific, MA, USA) with an IonPac AG14 Guard column and an AS14 analytical column with temperature of 30°C. The eluent was the mixture of 6 mM Na2CO3 and 5 mM NaHCO3 with a flow rate of 1.0 ml/min. Standard curves of anions whose correlation coefficients all exceeded 0.999 were established before the measurement of samples.

1H nuclear magnetic resonance (NMR) spectra were acquired using a 500-MHz Avance III spectrometer (Bruker, Germany) at room temperature. Generally, 0.25 ml of sample solutions was mixed with 0.05 ml of D2O (99.9%; Sigma-Aldrich) and was placed in an NMR tube with 5-mm outside diameter. A solvent suppression was run to minimize the solvent signal, and 5 mM 3-(trimethylsilyl)-1-propanesulfonic acid-d6 sodium (DSS-d6; Sigma-Aldrich) was used for the calibration of the 0 ppm in advance.

Evaluation of photovoltaic efficiency

The IPCE in the photochemical experiment under monochromatic light was estimated according to the yield of HCOOH, using Eq. 13.IPCE=nenp=2×NHCOOH×NA(Plight×t×A)/(hv)(13)where ne, np, NHCOOH, NA, Plight, t, A, h, and v are the number of electrons for HCOOH production from CO2, the number of incident photons, the mole number of produced HCOOH, Avogadro constant, the power of monochromatic light reaching reactor, irradiation time, the exposed area of reactor, Planck constant, and the frequency of monochromatic light, respectively.

Model description and the estimation of organic production

The model assumed that there were primitive carbonate terrestrial hot springs, whose area ratio (A, %) to the total surface area of early Earth (radius as 6371 km) was undetermined. Inside those springs, S0 particles in a certain concentration (ms, mM) were suspended in the UVC-penetrated photic zone with a certain thickness (DUV, cm). On the basis of the above assumptions, the volume (V, m3) of the photic zone in terrestrial hot springs is a function of A and DUV (Eq. 14). Considering the relationship between daily HCOOH production (Cd, mM day−1) and the concentration of S0 (Eq. 8 and fig. S5), the annual production of HCOOH (Cy, kg year−1) could be described as Eq. 15 (detailed form of Eq. 9) with respect to ms, A, and DUV.V=4π×(6371×103)2×DUV100×A=5.10×1012DUVA(14)Cy=V×Cd×365×46×103(15)


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: All the coauthors give their deep appreciation to Meixiang Zhu, who passed away during the revision of this work. Her contributions to this work were tremendous, in the absence of which our fieldwork would not have been possible. We thank X. Bai and Y. Chen for their dedications to the design and plotting of Figure 6. Funding: This work was supported by the National Natural Science Foundation of China (grant nos. 41872042, 91951114, 91851208, and 41522201). Author contributions: Yan Li and A.L. designed the study and conceived the idea for the paper. Yanzhang Li, Yan Li, Yi Liu, Y.W., A.L., J.W., H.J., and X.W. performed photochemical experiments. Yanzhang Li, B.W., and L.L. finished mineralogical and semiconducting characterization of samples. H.Y. performed bandgap DFT calculation. Yanzhang Li, Yan Li, A.L., and Yi Liu analyzed the data. M.Z., Yong Lai, Yan Li, Yanzhang Li, Y.W., and H.J. led the fieldwork and sample collection. Yanzhang Li and Yan Li wrote and revised the paper. J.D. helped polish the paper. Yan Li, H.D., Yong Lai, C.W., and A.L. discussed the focus of the work. All authors provided critical feedback on drafts of this paper and approved the final version. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article