A mineral-based origin of Earth’s initial hydrogen peroxide and molecular oxygen

Significance

Molecular oxygen (O₂) is essential for respiration on today’s Earth, while toxic to emerging anaerobic microbes or prebiotic chemistry in the Archean. Early life acquired a defensive ability against reactive oxygen species (ROS) in developing Archean oxic microenvironments. Detailed knowledge of coeval abiotic ROS sources is important for understanding the evolution of early life and planetary habitability. Mineral surfaces are known to produce ROS via splitting water. We experimentally find that ROS production at mineral–water interfaces derives oxygen from minerals as well. This reaction may be initiated by mechanical forces in various geodynamic processes, which deform minerals to produce surface radicals for releasing oxygen by interaction with water. Such rocky oxidants created opportunities for life and drove its early evolution.

Abstract

Terrestrial reactive oxygen species (ROS) may have played a central role in the formation of oxic environments and evolution of early life. The abiotic origin of ROS on the Archean Earth has been heavily studied, and ROS are conventionally thought to have originated from H₂O/CO₂ dissociation. Here, we report experiments that lead to a mineral-based source of oxygen, rather than water alone. The mechanism involves ROS generation at abraded mineral–water interfaces in various geodynamic processes (e.g., water currents and earthquakes) which are active where free electrons are created via open-shell electrons and point defects, high pressure, water/ice interactions, and combinations of these processes. The experiments reported here show that quartz or silicate minerals may produce reactive oxygen-containing sites (≡SiO•, ≡SiOO•) that initially emerge in cleaving Si–O bonds in silicates and generate ROS during contact with water. Experimental isotope-labeling experiments show that the hydroxylation of the peroxy radical (≡SiOO•) is the predominant pathway for H₂O₂ generation. This heterogeneous ROS production chemistry allows the transfer of oxygen atoms between water and rocks and alters their isotopic compositions. This process may be pervasive in the natural environment, and mineral-based production of H₂O₂ and accompanying O₂ could occur on Earth and potentially on other terrestrial planets, providing initial oxidants and free oxygen, and be a component in the evolution of life and planetary habitability.

Recent phylogenetic analyses (1–4) suggest that early life evolved in weakly oxic microenvironments within early Earth’s anoxic atmosphere [10^–5 ~ 10^–7 atm (5–9)]. Before the emergence of oxygenic photosynthesis, the oxic environments are thought to have been created and maintained by several proposed abiotic oxidant sources existing on the early Earth (10, 11). Many studies focus on the formation of reactive oxygen species [ROS, including hydroxyl radical (•OH), hydrogen peroxide (H₂O₂), superoxide radical (•O₂^–), and singlet oxygen (¹O₂)] in the atmosphere, such as photolysis of atmospheric water vapor (12–15), heterogeneous reactions on atmospheric aerosols (16–19), and the water–air interface of aqueous microdroplets (20, 21). Alternatively, ROS generated at mineral–water interfaces has drawn increased attention in recent years due to its high potential bioavailability on the Earth’s surface (22–31). Formation of oxic conditions requires ROS sources to overwhelm its sinks, but abundant reducing substances (H₂S, FeS₂, etc.) on the early Earth do not favor ROS accumulation (12, 23). As the most abundant minerals in Earth’s surface, silicates are resistant to oxidative weathering, especially for felsic silicates with fewer reductants, endowing them with a better potential for ROS local accumulation. Silicate–water interaction is therefore a promising solution to the ROS accumulation problem to provide a fundamental and persistent oxygen source on the Archean terrestrial surface.

ROS-producing reactions at silicate–water interfaces initiated by mechanical forces have been widely explored in mechanochemistry and medical science (32–36), which are potentially important in planetary chemistry as well. Even during modest rock friction, the strength at the actual contact interfaces could reach at least several giga pascal (37). Besides causing microscopic thermal weakening and melting of asperity tips (38), the mechanic force could readily break strong covalent Si–O bonds in [SiO₄]^4– tetrahedra [dissociation energy: 793.2 ± 16.4 kJ mol^–1 (39)] in silicate minerals (Fig. 1A), leading to the formation of surface-bound radicals (SBRs) with unpaired electron on the shell of crystal, such as E′ center (≡Si•), oxygen radical (≡SiO•), and peroxy radical (≡SiOO•) (34, 40) (Fig. 1B). It is the applied stress that provides the deformational energy to the potential energy surface, reducing the activation energy for bond scission in mineral crack propagation within strained crystals and ultimately increasing the probability of bond dissociation (41, 42). The mechanically induced process produces lattice defects and SBRs as gamma ray irradiation and thermal treatment also do (43, 44). The highly reactive SBRs interact with the surrounding water to generate hydrogen (H₂) (45), molecular oxygen (O₂), and ROS (32, 33, 46, 47) (Fig. 1 C and D). The hydroxylation of these defects due to their electron activity is energetically favorable (48–50).

Fig. 1.

Two proposed mechanisms for H₂O₂ generation at quartz–water interfaces. (A) Stress-induced cleavage of Si–O bonds. Applied force is represented by a linear falling potential (dashed line), which is added to the Morse potential of the unstretched Si–O bond (dotted line), thus resulting in a force-deformed Morse potential (solid line). The new dissociation energy D′ is smaller than D in the case of a stretching force. Modified after Kauzmann et al. (42). The probability of generation of surface-bound radicals in quartz therefore increases during mechanical processes. (B) Contact of water on freshly fractured surface of abraded quartz. Modified after Schoonen et al. (32). (C) Generation of H₂O₂ and O₂ in the reaction between H₂O and oxygen radical (≡SiO•). (D) Generation of rocky H₂O₂ and O₂ in the reaction between H₂O and peroxy radical (≡SiOO•). Surface-bound radicals are formed through homolysis of Si–O bonds, including ≡SiO•, ≡SiOO•, and E′ center (≡Si•), while surface charges (≡Si⁺ and ≡SiO^–) for heterolysis. Silanol groups (≡SiOH) are end products on quartz surface after hydroxylation.

Stress-induced ROS generation at silicate–water interfaces in several geological processes has been investigated (25, 26, 51). Experiments mimicking abrasion of quartz in turbulent subaqueous environments indicate that a significant amount of H₂O₂ could be produced, creating and sustaining oxic microenvironments in Archean deltas and seashores (25). Following the SBR-induced ROS formation, it is proposed that rock faulting in geothermal subsurface environments can lead to H₂O₂ formation (51). However, a deep mechanistic ROS formation reaction framework of silicate–water interaction remains lacking, hindering further exploration of how general geological processes triggered the release of ROS to the early Earth at a global scale and drove the evolution of ROS-detoxifying system and oxygenic photosynthesis. For example, relatively small earthquakes in the tectonically inactive crust of early Earth are sufficiently strong to allow rock fracturing (30, 31). Surface and ground water readily flow downward into cracks and trigger water–rock reactions such as serpentinization that provide chemicals (e.g., H₂ and CH₄) and energy to sustain a potentially habitable subsurface environment on the early Earth. In reaction sequences such as this, the role of other water–rock reactions and products such as H₂O₂ formation should be experimentally and theoretically explored. Here, we highly focused on the experiments that unambiguously elucidate several hypothesized silicate-induced ROS formation mechanisms. This work allows for a quantification of potential oxygen sources and provides further mineralogical evidence that oxidative stress coming from early minerals promoted the evolution of life on Earth. This is critical for understanding the intrinsic links in coevolution of the lithosphere, hydrosphere, atmosphere, and biosphere (52–55).

Possible Mechanisms for H₂O₂ Production at Abraded Mineral–Water Interfaces

≡SiO• is one type of initial oxygen-bearing radicals on freshly fractured surfaces of abraded quartz, which is formed through homolysis of Si–O bonds in ≡Si–O–Si≡ (Eq. 1) (34, 35) or O–O bonds in peroxy linkages (≡Si–O–O–Si≡) (45) (Eq. 2) (Fig. 1B), with an oxide O^2– ion trapping a hole. It is generally assumed that •OH is first produced via water splitting by ≡SiO• (hydrogen abstraction) (Eq. 3) (36, 47), and •OH further recombines to H₂O₂ (Eq. 4) (Fig. 1C) (26, 51, 56, 57). In this case, the oxygen atoms in •OH and H₂O₂ ultimately originate from water (Fig. 1C). The complete reaction sequence for H₂O₂ generation, initiating from the formation of ≡SiO• on a surface, either geological or aerosol with liquid layer, by

$$\begin{array}{c}\equiv \text{Si}–\text{ O }–\text{Si}\equiv \to \equiv \text{SiO}•+\equiv \text{Si}•,\hfill \end{array}$$ [1]

$$\equiv \text{Si}–\text{ O }–\text{ O }–\text{Si}\equiv \to 2\equiv \text{SiO}•,$$ [2]

$$\begin{array}{c}\equiv \text{SiO}•+{\text{ H}}_{\text{2}}\text{ O}\to \equiv \text{Si}–\text{OH}+•\text{OH,}\hfill \end{array}$$ [3]

$$2•\text{OH}\to {\text{ H}}_{\text{2}}{\mathrm{\text{O}}}_{\text{2}}\text{.}$$ [4]

Alternatively, H₂O₂ is a possible final product of reactions between water and peroxy radical (≡SiOO•) (Eqs. 5–7) wherein both H₂O₂ and ≡SiOO• contain a peroxy bond (O–O). As mentioned earlier, ≡SiOO• could directly form in homolysis of ≡Si–O–O–Si≡ (Eq. 5) (Fig. 1B) (25, 48), which is produced via dislocation slip during plastic deformation (58) or condensation of hydroxyl pairs (26), is readily preserved in clastic, igneous, and high-grade metamorphic rocks, and has been widely observed in silicates (26, 44, 58). The O–O bond is generally weak and may therefore dissociate via the interaction of H₂O. It is also possible that both oxygen atoms in ≡SiOO• transfer to H₂O₂ because the free-energy barrier of this route may be significantly reduced by catalysis such as water or acids, as observed in experimental and theoretical studies of organic hydroperoxide chemistry (59–61). In either case, the oxygen atoms in peroxy radicals, rather than water molecules alone, could have contributed to HO₂• (perhydroxyl radical), H₂O₂, and O₂ production (Fig. 1D). A most recent experiment observed oxygen transfer from a ¹⁸OH-labeled SiO₂ substrate to H₂O₂, but the exact mechanism remains elusive (62). We propose that ≡SiOO• plays a crucial role in oxygen transfer to H₂O₂ and that oxygen in water may transfer to a silicate surface (Fig. 1D), which has not yet been demonstrated and forms the basis for our experiments. Of particular importance and application is that the subsequent disproportionation of H₂O₂ to O₂ implies that the molecular oxygen on Earth, especially before the advent of oxygenic photosynthesis, can originate from silicate rocks. The potential oxygen transfer between silicate rocks, water, and ROS may provide a unique isotope proxy to examine this intriguing but overlooked process in the geological past because ROS is so reactive that can hardly be preserved in geological archives, especially those from billion years ago. The process of Eq. 6 is particularly critical to justify and its rate and mechanism must be thoroughly known.

$$\begin{array}{c}\equiv \text{Si}–\text{ O }–\text{ O }–\text{Si}\equiv \to \equiv \text{SiOO}•+\equiv \text{Si}•,\hfill \end{array}$$ [5]

$$\begin{array}{c}\equiv \text{SiOO}•+{\text{ H}}_{\text{2}}\text{ O}\to \equiv \text{Si}–\text{OH}+{\text{HO}}_{\text{2}}•,\hfill \end{array}$$ [6]

$$2{\text{ HO}}_{\text{2}}•\to {\text{ H}}_{\text{2}}{\text{ O}}_{\text{2}}+{\text{ O}}_{\text{2}}\text{.}$$ [7]

Here, we report a series of ¹⁸O-labeling experiments (Materials and Methods) to define the oxygen-generating mechanism (Fig. 1D). By this pathway, oxygen atoms in quartz (or silicate minerals in general) detach from mineral surfaces speciated as ROS and O₂ when reacting with water; concomitantly, the oxygen from the aqueous media attaches to mineral surfaces via hydroxyl formation. Given that ≡Si–O–O–Si≡ linked to a wide range of geological processes is present in most silicates, H₂O₂ production associated with ≡SiOO•, if demonstrated, ostensibly could be a widespread phenomenon and occur in ubiquitous environments. Notably, the formation of peroxy linkage is not restricted to silicates but is known to occur in oxides and other minerals (63). In the present experiments, quartz is utilized as a model mineral as it is a primary rock-forming mineral, structurally similar to tectosilicates, and the simplest silica mineral with bridging oxygen (BO, e.g., ≡Si–O–Si≡ and ≡Si–O–O–Si≡) in [SiO₄]^4– tetrahedra that provide the provenance for SBRs via homolytic cleavage. Our experiments were performed under strictly anoxic conditions to mimic early Earth and other planetary environments (Materials and Methods).

Experiments on Stress-Induced Reactive Sites on Abraded Quartz Surface

We first confirm the emergence of reactive oxygen sites on the abraded surface of a quartz sample (QGN) that was ground in a pure, dry nitrogen atmosphere (Materials and Methods). The nonbridging oxygen (NBO), including SBRs (≡SiO• and ≡SiOO•) and broken bonds (≡SiO^–), appear on the abraded quartz surface (25, 34), as shown by the distinct changes in the X-ray photoelectron spectroscopy (XPS) spectra (Fig. 2). In the XPS spectra of O 1s, O 2s, and Si 2p (Fig. 2 A–C), BO contributes to the bulk of these bands at 532.7, 103.4, and 25.4 eV, respectively. In contrast to the intact quartz (QW), the little band dispersion appears in the abraded quartz. Accordingly, the changes in combination state of [SiO₄]^4– tetrahedra show up in the valence band (VB, 0 to 20 eV) spectra (Fig. 2 C and D). The rupture of BO in Si–O σ-bonds in [SiO₄]^4– tetrahedra causes a strong spectral loss in the lower VB (10.35 to 16.60 eV), while the produced NBO leads to a spectral increase in the upper VB (10.35–ca. 5 eV). The appearance of ≡SiOO• in QGN is also supported by electron paramagnetic resonance (EPR) measurements (SI Appendix, Fig. S1A). Compared with the aged quartz (QGN-H₂O), it can be found that a considerable proportion of ≡SiOO• preserved on the surface of QGN, which contribute to the surface reactivity of QGN (25, 63).

Fig. 2.

XPS evidence for the formation of reactive sites on abraded quartz surface. (A and B) O 1s and Si 2p spectra of abraded quartz and intact quartz. The fitted peaks for the spectra of abraded quartz correspond to bridging oxygen (BO, i.e., ≡Si–O–Si≡ and ≡Si–O–O–Si≡) and nonbridging oxygen (NBO, i.e., ≡SiO^–, ≡SiO•, ≡SiOO•). (C) Comparison between the valence band and O 2s spectra of abraded quartz and intact quartz. The valence band spectrum is assigned to NBO orbitals (1e, 5t₂, 1t₁ orbitals) of mostly O 2p_y character and BO σ-bonding orbitals (5α₁, 4t₂) of mostly Si sp³–O 2p_z character. The O 2s band consists of a main peak (3t₂) and a shoulder (4a₁). (D) A differential spectrum obtained by subtracting the spectrum of intact quartz from that of abraded quartz.

Such NBO sites make the surface of the abraded quartz much more reactive than that of intact quartz. As Fig. 2D shows, the band of the highest occupied molecular orbital (HOMO) shifts upward from 4.92 eV (intact quartz) to 4.47 eV (abraded quartz), indicating that the electrons in the O 2p orbitals become more energetic, which is consistent with the calculated changes in the electron density (48). These reactive oxygen sites can readily react with H₂O, leading to a significant decrease in surface energy (64), and their lifetime mostly depends on the rate of water delivered to the abraded surface. We experimentally monitored the interaction between water and the NBO sites and the time course of hydroxyl species formation on the abraded quartz surface, and it is shown in SI Appendix, Fig. S2. Hydroxylation of the pristine surface began with the formation of isolated (≡SiOH) and geminal silanol (=Si(OH)₂) groups, followed by transformation into hydrogen bonds interacting hydroxyl groups (e.g., ≡SiOH…OH), and terminated with the establishment of a complex hydrogen bond network between water molecules (SI Appendix). During the interaction between SBRs and water molecules at atomic scale lengths, the ROS generation is completed within 1 μs (50).

A Mineral Oxygen Origin of ROS Generated at Quartz–Water Interfaces

Based on the knowledge of the surface reactivity of the abraded quartz (QGN) and its behavior when interacting with water, we employed ¹⁸O-labeled water (H₂¹⁸O, at. 97%) as a tracer to track the oxygen source of ROS (Fig. 3A), with ¹⁸O in the abraded quartz at natural abundance levels (~0.2%). In the reactions between water and QGN, •OH, H₂O₂, and O₂ are formed simultaneously (SI Appendix, Table S3), while the counterpart (•H) (Eq. 8) combines to form H₂ (Eq. 9) that escapes the system (25, 45). We used a benzoic acid (BA) method to measure the isotope composition of •OH (Materials and Methods). Benzoic acid (BA) can react with •OH at a fast rate (k = 4.3 × 10⁹ M^–1 s^–1) to form p-hydroxybenzoic acid (p-HBA) (Eq. 10). Thus, the p-HBA formed with •¹⁶OH [p-HBA(¹⁶OH)] or •¹⁸OH [p-HBA(¹⁸OH)] has different m/z ratios (137 and 139, respectively) in mass spectrometric measurements (65). We do not observe any significant isotopic exchange between BA and ¹⁸O-labeled water that may confiscate our measurements (Materials and Methods and SI Appendix, Fig. S4). Consequently, the reaction between •OH and BA, a low-weight organic acid, is tracked by the labeling technique and in general mimics the interaction between ROS and organic molecules.

$$\equiv \text{Si}•+{\text{ H}}_{\text{2}}\text{ O }\to \equiv \text{Si}–\text{OH}+\text{ H}•,$$ [8]

$$2\text{ H}•\to {\text{ H}}_{\text{2}},$$ [9]

Fig. 3.

The generation and content of ¹⁸O labeled p-HBA in the ¹⁸O labeling experiments. (A) The scheme of the ¹⁸O-labeling experiments [Set I: QGN+BA(H₂¹⁸O); Set II: QGN+H₂¹⁸O; Set III: QGO+BA(H₂¹⁸O); Set IV: QGO+H₂¹⁸O]. (C) ROS production in the suspension of QGN. (B and D) The content of p-HBA(¹⁶OH) and p-HBA(¹⁸OH) that produced after QGN/QGO was added into BA(H₂¹⁸O) solution or H₂¹⁸O (Set I–IV experiments), respectively. The O_Total is the total O atoms from •OH and H₂O₂. The error bars present ±1 SD of three independent replicates.

$$•\text{OH}+\text{ BA}\to p-\text{HBA.}$$ [10]

If the oxygen atoms in ROS exclusively originate from water (Eqs. 1–4), the formation of ¹⁸O-rich ROS and subsequent p-HBA(¹⁸OH) is expected. Surprisingly, when the QGN was added into the BA-H₂¹⁸O solution, both p-HBA(¹⁸OH) and p-HBA(¹⁶OH) were simultaneously produced from the oxidation of BA by •OH, which occurs at the quartz–water interface [Set I: QGN + BA(H₂¹⁸O)] (Fig. 3B, Left). Quantitative analysis (SI Appendix, Table S4) shows that the •OH mainly (~81%) stems from H₂¹⁸O splitting (with the product of •¹⁸OH), while the rest (~19%) derives from an oxygen reservoir with the ¹⁶O-dominant isotopic composition, i.e., the abraded quartz. Given that the reaction occurs at the quartz–water interface, the isotope transfer likely occurs on the surface of the abraded quartz. Surprisingly, in the p-HBA formed during the oxidation of BA by the newly produced •OH after the dissociation of the H₂O₂ in Fenton reactions (Materials and Methods), the p-HBA(¹⁸OH) accounts for only ~27% while the p-HBA(¹⁶OH) becomes the majority (~73%). The high ¹⁶O content in H₂O₂ is best explained by the Eqs. 6 and 7, where oxygen atoms in ≡SiOO• are involved in H₂O₂ formation. In Eq. 6, there is at least one oxygen atom from ≡SiOO• that transfers to HO₂•. If the bond dissociation in ≡Si–O–O• occurs at the O–O bond, there is only one oxygen in HO₂• derives from ≡SiOO• and the ¹⁶O contents of the H₂O₂ product are expected to be ~50%. Indeed, the high ¹⁶O content of H₂O₂ (>50%) indicates that both oxygen atoms in ≡SiOO• transfer to H₂O₂, and HO₂• is formed via the dissociation of the Si–O bond in ≡Si–O–O• (Fig. 1B). Since the ¹⁸O-enriched •OH was quickly consumed by BA in the solution (Eq. 10) (Fig. 3B, Left), the formation of H₂O₂ via the combination of •OH (Eq. 4) produced from ≡SiO• sites (Eqs. 1 and 2) plays a minor role in our experimental results. The abundant p-HBA(¹⁶OH) observed in our ¹⁸O-labeling experiments therefore challenges the conventional view that the oxygen atoms in ROS originate exclusively from water splitting. Alternatively, our findings indicate that it is the oxygen atoms in the SBRs that mainly contributed to the formation of H₂O₂ and support our hypothesis that ≡SiOO• is the dominant precursor site for H₂O₂ production (Eq. 6) (Fig. 1D). Because H₂O₂ can be decomposed to •OH, this H₂O₂ production mechanism explains why the contribution of the oxygen from solid to •OH (~20%) is much higher than a most recent experiment using SiO₂ with ¹⁸OH-labeled surface (~1%) (62). In addition, because the production amount of •OH is much lower than H₂O₂, ~68% of the oxygen atoms in ROS measured here (•OH and H₂O₂) (O_Total) originate from the quartz surface, while only ~32% of O_Total is from water (Fig. 3C and SI Appendix, Table S4). This result unambiguously indicates a rocky origin of oxygen.

To further evaluate the role of the •OH recombination reaction in the H₂O₂ formation (Eq. 4), we also measured the oxygen isotopic composition of the H₂O₂ generated in the absence of organic molecules (i.e., BA here) during the ROS generation at quartz–water interfaces (Set II: QGN + H₂¹⁸O). After the formation of H₂O₂ in the BA-free H₂¹⁸O with QGN, BA and Fenton reaction reagents were introduced and the isotopic compositions of the H₂O₂ were measured (SI Appendix). In the produced p-HBA, p-HBA(¹⁶OH) accounts for ~80% of the total p-HBA amount (Fig. 3B, Right), similar to experiments with BA (Fig. 3B, Left). This result further supports the contention that a large proportion of the oxygen in the H₂O₂ originated from ≡SiOO• sites over the abraded quartz surface irrespective of reaction conditions. Given that the ¹⁸O-rich signature is not found in H₂O₂, •¹⁸OH formed by water (H₂¹⁸O) splitting is likely consumed by interaction with HO₂• (an intermediate radical before H₂O₂, Eq. 7) (Eqs. 11 and 12, k = 1.0 × 10¹⁰ M^–1 s^–1) or H₂O₂ (Eq. 13). The relatively low H₂O₂ production in the absence of BA (Fig. 3C) supports our interpretation that H₂O₂ was destroyed by •OH.

$$•\text{OH}+{\text{ HO}}_{\text{2}}•\to {\mathrm{\text{H}}}_{\text{2}}\mathrm{\text{O}}+{\text{ O}}_{\text{2}}\text{,}$$ [11]

$$•\text{ H}+{\text{ O}}_{\text{2}}\to {\text{HO}}_{\text{2}}•,$$ [12]

$$•\text{OH}+{\text{ H}}_{\text{2}}{\mathrm{\text{O}}}_{\text{2}}\to {\text{ HO}}_{\text{2}}•+{\text{ H}}_{\text{2}}\text{O.}$$ [13]

We also verified the ≡SiOO•-site-associated mechanism for H₂O₂ production by carrying out Set III–IV experiments, in which quartz was ground in an O₂ atmosphere (QGO) so that a higher density of ≡SiOO• (25) in QGO than QGN is obtained via the interaction between O₂ and ≡Si• (Eq. 14) (SI Appendix, Fig. S1B). The H₂O₂ yield per unit surface area in QGO experiments is ~6.8-fold higher than that in QGN experiments (SI Appendix, Table S3), confirming the role of ≡SiOO• in promoting H₂O₂ production. Given that isotopic composition of both O₂ and quartz is in natural abundance (¹⁶O: ~99.75%), the enhanced formation of H₂O₂ via interaction between H₂¹⁸O and ≡SiOO• sites on QGO is expected to result in a higher ¹⁶O content in H₂O₂ than QGN. This prediction is perfectly supported by the content of p-HBA(^16/18OH) produced in the suspension of QGO (Fig. 3D). The percentage of ¹⁶O in ROS is generally higher than that of QGN, especially for the H₂O₂ in the QGO+H₂¹⁸O experiment (¹⁶O = ~90%). We note that O₂ produced from ROS in the QGN suspension could return to the quartz surface by forming ≡SiOO• (Eq. 14) or bonding with •H (Eq. 12) (25). These reactions require that there is a formation of a complex reaction network for H₂O₂ production in water (Eqs. 6 and 7) and oxygen transfer among solid, liquid, and gas.

$$\equiv \text{Si}•+{\text{ O}}_{\text{2}}\to \equiv \text{SiOO}•.$$ [14]

Oxygen Transfer Pathways between Silicate and Water via ROS Generation and Conversion.

Our ¹⁸O labeling experiments clearly demonstrate the facility of oxygen transfer from solids to liquid phases, which involve various ROS production and transformation processes. Because the oxygen atoms in H₂O₂ are predominately from ≡SiOO• on the abraded quartz surface, the oxygen transfer processes likely occur in the transition state. We propose the following mechanism: When the Si–O bond in ≡Si–O–O• is broken, the abstraction of the hydrogen atom from the water leads to the production of HO₂•, and a new Si–O bond (≡Si–OH) is formed between the Si atom and the hydroxyl group of the water (Eq. 6) (Scheme 1). In this case, oxygen transfers from water to solid, and ¹⁸O-enrichments in reacted quartz are expected in our experiments. Two different measurements (mass spectrometry analysis of thermally desorbed species from the quartz surface and direct isotopic analysis of the bulk quartz) reveal anomalous ¹⁸O enrichments in the quartz samples reacted with ¹⁸O-labeled water (as per SI Appendix, Fig. S5 and Table S6), confirming our proposed mechanism (Fig. 1D). Given that the oxygen transfer occurs at the surface, the isotopic composition of quartz after reactions becomes heterogenous. This prediction may be tested by high-spatial-resolution isotope measurements in the future, which is challenging and beyond the scope of this study. Notably, the oxygen-transfer pathways at the quartz–water interfaces are unexpected as isotopic exchange between water and a quartz surface is conventionally considered to occur between the dissolved SiO₂ species and water (66). Although oxygen exchanges in ROS aqueous reactions are extremely fast (10⁶~10¹⁰ M^–1 s^–1, 25°C) (67, 68), the rates of oxygen transfers are limited by the ROS-producing reactions at the quartz-water interfaces (2.3 × 10⁻⁶ M m^–2 s^–1, 25°C) (25), which is still much faster than oxygen diffusion in crystalline quartz under hydrothermal conditions (250 to 750°C) (69).

Scheme 1.

The proposed oxygen transfer pathways in the hydroxylation of the peroxy radical (≡SiOO•). The H atom of water binds with the terminal oxygen of surface peroxy radical (≡SiOO•), while the O atom of water attaches to the Si atom of ≡SiOO•, so that peroxy oxygen pair (O–O) remains intact and the HO₂• (perhydroxyl radical) is formed. Red letters indicate the rocky oxygen from silicates.

Oxygen transfer between SBRs and water reveals a ubiquitous geochemical pathway for rapid cycling of oxygen across Earth’s multispheres since such SBR formation and subsequent spontaneous ROS-producing processes occur not only in the crushing of quartz but also in natural silicates (25, 32, 46). Unlike photochemical processes that require ultraviolet photons to split water to •OH (13, 20), when silicate rocks are mechanically cracked and eroded by ubiquitous geodynamics, H₂O₂ and •OH are produced as soon as ≡SiOO• and ≡SiO• on the fresh surface are exposed to water, and most oxygen atoms in ROS originate from the rock rather than the water (Figs. 1D and 4). Concomitantly, the directly produced O₂ escapes to the atmosphere, and the H₂O₂ dissociated to O₂. The thermodynamically favorable process is comparable to recent findings of spontaneous H₂O₂ production in water microdroplets via hydroxide ions (20) but with a higher oxygen exchange capacity.

Fig. 4.

The rocky oxygen-driving evolution of ROS-detoxifying system in early life in the Archean. (A) ROS production in geological processes (e.g., earthquakes and river erosion). (B) Oxygen transfers during the ROS generation and conversion at interface between H₂¹⁸O and abraded quartz and ingestion by microorganisms. Early microorganisms coped with ROS by antioxidant enzymes. For example, the H₂O₂ is dissociated into O₂ by catalase. The red ball is ¹⁶O on quartz surface, and the pink ball is ¹⁸O that derives from H₂¹⁸O. For detail equations, see SI Appendix, Table S7.

With proper geologic proxies that can record isotope compositions of O₂, ozone, and radical-derived species (e.g., sulfate), our finding provides a prototype to examine various early Earth’s H₂O₂ production models (e.g., photochemical and mechanochemical). Given wavelength-dependent isotope effects (e.g., intersystem crossing) linked to highly structured cross sections of many atmospherically relevant molecules (e.g., CO, CO₂, and H₂O) at UV ranges, photolytic reactions usually lead to huge isotope anomalies (several hundreds and even thousands ‰) in ROS products (70). Alternatively, ROS in mechanochemical processes mainly inherits oxygen isotopic compositions of rocks (or with relatively small isotope fractionation), the isotope composition of such O₂ may be different from those generated from water or CO₂ photolysis. On geologic timescales, this oxygen transfer in the physical erosion of silicate rocks and subsequent sedimentary process might have produced a detectable geochemical record given the lithologic source. Although the isotope fractionation factors in the radical reactions at quartz-water interfaces could not be determined in this study, we highlight that such rapid oxygen transfer between water molecules and SBRs on abraded quartz surface is a hitherto unrecognized component that has broad implications and deserves further investigation in the future. One example of importance is the secular changes of oxygen isotope compositions in cherts and other sedimentary rocks (e.g., carbonate and shale) across the Earth’s history, which have been interpreted as a hot (up to 70°C) Archean Ocean, changes of seawater oxygen isotope compositions, progressive diagenetic alteration, or their combinations (66, 71–73). It is preliminary to state that this is correct, but the experiments clearly suggest that modeling of these systems would be of high value. The oxygen transfer between water and quartz via ROS production is a kinetic process different from isotopic exchange equilibria. The kinetic isotope effect remains undefined but needs to be quantified in the future and properly corrected for as this process may alter the original isotope composition of chert deposits and potentially influence the oxygen isotope composition of seawater (72). In addition, this oxygen transfer mechanism also has implications for relevant extraterrestrial environments. Oxygen transfer among atmospheric oxygen-bearing molecules (e.g., CO₂, O₂, and O₃), thin water layers, and carbonate minerals surface has been documented and proposed as a probe to examine Martian meteorite (ALH84001) isotope data for potential testing of “warm and wet” and “cool and dry” models of early Mars (74, 75). Anomalous isotopic effects in SiO₂ formation (SiO + •OH, a reaction relevant to this study) have been experimentally demonstrated to explain the oxygen isotope distribution of our solar system (76). The surface-induced isotope exchange mechanism discovered in this study could therefore complement these interpretations as noted by a recent review (70).

Role of Rocky Oxygen in the Evolution of Early Life and Planetary Habitability.

The mechanism of mechanochemical ROS formation defined in this study suggests that this abiotic ROS-producing process may have a profound effect on the evolution of early life. In particular, abiotic productions of H₂O₂ and O₂ have long been proposed as a crucial step that enabled the evolution of ROS-detoxifying and oxygen-utilizing enzymes and paved the way for the emergence of aerobic metabolism and oxygenic photosynthesis (10, 11, 25, 77, 78), while the highly reactive •OH directly damages cells (79). In conventional views, H₂O₂ production via all water-splitting pathways involved the formation of •OH as a precursor (Eq. 4). However, our experiments show that •OH is not required because H₂O₂ and O₂ are readily produced from ≡SiOO• via the HO₂• recombination reaction rather than ≡SiO• via the •OH recombination reactions. As the present BA experiments (Fig. 3C) suggest, in the habitats of microorganisms where organic molecules [e.g., dissolved humic substances (80) and extracellular polymeric substances (81)] are common, H₂O₂ readily accumulates to micromolar levels (25) because •OH is readily consumed by organic molecules and the decomposition of H₂O₂ by •OH is therefore hindered. Thus, we argue that in aqueous environments with intensive mechanical processes upon silicate rocks, peroxy linkages (≡Si–O–O–Si≡) and peroxide groups (≡SiOO•) are more likely to be formed for H₂O₂ production, which is relatively available for early life (Fig. 4).

A key geological constraint on the oxidative adaptation for early life is the wide formation of weakly oxic microenvironments on the Archean Earth (1, 11), and rock friction by various geodynamics can provide a substantial and persistent oxygen source. An experimentally based estimation suggests that on the early surface environments where quartz was abraded by coastal waves and tides, H₂O₂ could be generated at the rate of 4.87 × 10¹¹ molecules cm⁻² s⁻¹ (25). Recent studies suggest that tectonic activities could also provide substantial H₂O₂ in the modern Earth. For example, rock crushing in subglacial or geothermal regions provides sufficient H₂ and H₂O₂ to sustain subsurface microbial ecosystems (57). Here, we extend this discussion to the early subsurface crust. On the early Earth, the seismic activity could have actively proceeded in early subsurface crust during fault movements associated with vertical processes of Archean cratons before the onset of plate tectonics (82). The earthquake magnitude (M) of these events is generally small but sufficiently large to allow for faulting and subsequent rock–water reactions. Previous studies estimate the cumulative flux of seismic H₂ (Eqs. 8 and 9) on fault surface attributed to earthquake to be 2.3 × 10⁵ mol m⁻² yr⁻¹ (M = 0 ~ 4) (83). This estimation is based on the number of ≡Si• generated from earthquake faulting (Eqs. 8 and 9), and ROS could be produced by an equal number of oxygen-bearing radicals (≡SiO• and ≡SiOO•) (25, 32, 44). Because the product H₂O₂ accounts for 97% of ROS (SI Appendix, Table S3, 3% for •OH), a seismic H₂O₂ flux is estimated to be 2.2 × 10⁵ mol m⁻² yr⁻¹ or 4.2 × 10¹⁷ molecules cm⁻² s⁻¹ based on our experimental results. The seismic H₂O₂ flux is 11 magnitudes higher than the Archean atmospheric photochemical H₂O₂ flux (~10⁶ molecules cm⁻² s⁻¹) (13). Such earthquakes and associated mineral–water reactions could weakly oxidize the fluids that circulate through fault zones and overflow to the terrestrial surface, as observed at the Sohna (Haryana, India) thermal spring water with H₂O₂ concentration in a range of 0.53 to 0.56 mg L^–1 or 15.59 to 16.47 μM (84). Importantly, the resultant H₂O₂ levels are high enough for early life to gradually develop a defense against oxygen toxicity (25, 85). We note that this production pathway of rocky oxygen in the crust is renewable in geological time because surface radicals (e.g., ≡SiOO•) can be easily regenerated by constant rock faulting induced by small intraplate earthquakes (30). We note that quartz and tectosilicates (e.g., feldspar) possess more bridge oxygen than other silicates for homolysis cleavage, and therefore, quartz- and feldspar-rich granitoids are expected to be the major minerals that produce significant amounts of ROS in mechanical processes. ROS production was likely spatially widespread, not only in fault zone channels where quartz could form when hydrothermal water cools (86) but also in stressed and crushed granites that intruded in early cratons (26). We envision that once substantial granitoids formed in the crust and global plate tectonics initiated in the Archean (87), aggressive tectonics and physical weathering of the felsic rocks could have led to a higher ROS-producing flux.

When the habitats of early microorganisms [e.g., benthic microbial mats in waters (25, 88) and methanogens in upland soils (51)] were oxygenated at microsites (89, 90) or at global scales (7, 91, 92) on ancient Earth, antioxidant enzymes, including peroxidases, catalases (CATs), superoxide reductases (SORs), and cytochrome oxidase, could help the anaerobic microbes to cope with oxidative stress (93–95). Phylogenetic evidence suggests that the ancestors of cyanobacteria used CATs and SODs to remove H₂O₂ and its derivatives, superoxide radical (•O₂^–), at least 3.3 to 3.6 billion years ago (79), well before the first rise of atmospheric oxygen during the Great Oxidation Event (ca. 2.45 billion years ago) (7). In addition, a recent study (80) also demonstrates that cytochrome c peroxidases in anaerobes enable the use of abiotic H₂O₂ as a terminal electron acceptor for anaerobic respiration under anoxic conditions, simultaneously defending the cell against its toxicity.

Conclusions

Our study delineates the chemical mechanism of O₂/ROS production at silicate–water interfaces and unambiguously demonstrates that oxygen atoms in O₂ and ROS are from the lithosphere. Minerals could convert mechanical energy into chemical energy and store via forming defects and radicals, and then provide the chemical energy for life via releasing ROS in mineral–water reactions. On the Archean Earth, under the stimulation of the rocky oxidants, early life might have evolved to handle O₂/ROS-involving reactions and aerobic respiration (3, 78), and these ROS-detoxifying pathways enabled them to survive when the environmental O₂ levels abruptly increased (7). Even on other terrestrial planets or icy satellites, rock fracturing actively occurs in various geodynamic processes, such as frequent sandstorms and quakes on Mars and strong tidal strains on Enceladus (31, 96), and the lithic oxygen production is physically possible to occur and leads to environmental O₂ fluctuation (56, 97). This ubiquitous abiotic oxygen source can be important for the detection of extraterrestrial life, as highly energetic aerobic respiration and even the evolution of complex multicellular life are possible (98). Future oxygen yield experiments are needed in order to quantify the relative roles of different processes and how much oxygen may be introduced to the environment over geologic time scales.

Materials and Methods

Sample Preparation and Storage.

Quartz was purchased from Richjoint (Shanghai, China). The 0.25 to 0.6-mm quartz particles were chosen and washed with deionized water for a few times until the pH of the cleaning solution reached neutral. Then, they were dried in an oven at 110°C for 24 h and kept in a glove box for further experiments (N₂ > 99.999%, H₂O < 0.1 ppm, O₂ < 0.1 ppm, Mikrouna).

To mimic the mechanical erosion of rocks in the Archean anoxic atmosphere, the quartz particles were milled in a N₂ atmosphere (O₂ < 0.1 ppm). In the glove box, the quartz sand (0.25 to 0.6 mm) was loaded into a zirconia ceramic jar with two inlet switches and sealed airtight with a gasket ring and screw fasteners. The jar was filled with the same atmosphere as in the glove box (i.e., ultrapure nitrogen). Subsequently, the jar was transferred to a planetary ball mill (Fritsch, Pulverisette 6) for grinding at 350 rpm for 3 h. After grinding, the jar was moved back to the glove box, and the abraded quartz samples (QGN) with a specific surface area of 4.35 m² g^–1 were collected and sealed in glass bottles to protect the surface reactive sites. Then, a part of QGN was washed with hydrofluoric acid (10 wt.%) for 8 h and dried in an oven at 60°C to remove SBRs and the amorphous layer in QGN, obtaining intact quartz (QW) for comparison. In addition, a part of QGN was washed with deionized water for five times and dried in an oven at 60°C, obtaining an aged quartz sample (QGN-H₂O) without surface radicals but with inner radicals in the amorphous layer.

The sample that was ground in a pure O₂ atmosphere for 5 h (QGO) was prepared by the following procedure: First, the quartz sand was loaded into the zirconia ceramic jar in the glove box, and the jar was sealed tightly. Then, the jar was removed from the glove box and vacuumed through an inlet with a mechanical pump. The jar was filled with pure oxygen (>99.99 %) before being transferred to the planetary ball mill for grinding at 350 rpm for 5 h. Finally, the sample with a specific surface area of 8.44 m² g^–1 was transferred to glass bottles in a desiccator filled with ultrapure N₂. This sample was used for the corresponding experiments within 48 h after grinding.

Characterization of the Quartz.

Surface-bound radicals created via grinding were measured by electron paramagnetic resonance (EPR) on a Bruker A300-10-12 spectrometer. The settings for the EPR measurements were as follows: center field, 3320 Gauss; sweep width, 500 Gauss; microwave frequency, 9.297 GHz; modulation frequency, 100 kHz; power, 2.28 mW; and temperature, 77 K.

Although dangling bonds are sensitive paramagnetic centers for EPR signals, they distribute not only on the abraded quartz surface but also in the amorphous layer (25, 34), even in crystal defects (44). For example, there is a considerable amount of radicals in aged quartz even though it has low surface reactivity (34). Here, the surface reactivity of abraded quartz can be evaluated via analyzing the electronic structure of the abraded quartz surface, which can be depicted from XPS spectra recorded from the quartz surface at a shallow depth of 1.6 nm, especially from the chemically sensitive valence-band spectra. XPS measurements were performed on a Thermo Scientific K-Alpha XPS instrument equipped with an Al Kα source (1,486.8 eV) operated at an emission current of 3 mA and a tube voltage of 12 kV. The vacuum in the analysis chamber was <1 × 10^–8 mbar. To prevent the charged surface interfering with the analysis, the electron neutralizer (flood gun) in the analysis chamber was activated during the analysis processes. The core levels (O 1s, Si 2p, and C 1s lines) were collected with a pass energy of 30 and a step size of 0.05 eV. Valance band (VB) spectra were collected at a pass energy of 50 eV with a step size of 0.2 eV. The VB spectra were acquired over a binding energy range of 34 eV, and each spectrum was acquired in 60 sweeps. At least three separate VB spectra were collected for each sample to test reproducibility and to monitor changes during exposure to the X-ray beam. XPS spectra were calibrated with the C 1s peak of adventitious carbon contamination at 284.8 eV. The spectra were processed using the Thermo Avantage analysis software with regard for the smart background and the mixed Lorentz–Gaussian shapes of the peaks.

In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed using a Bruker Vertex-70 infrared spectrometer over 128 scans at a resolution of 4 cm⁻¹. The reflectance spectra of samples were ratioed against an Al mirror background reference. To avoid the contact between fresh quartz surface and H₂O in air, QGN was loaded into an 8-mm diameter microsample cup in the glove box. The microsample cup was sealed in sample chamber before being removed from the glove box, and the sample was separated from the air during the measurements. After a stable signal was obtained from the fresh quartz surface, the valve was open and moist nitrogen at the humidity of about 75% was pumped into the sample chamber. All the scans were performed at room temperature (approx. 25°C).

The ¹⁸O/¹⁶O ratios of bulk quartz before and after reacting with the ¹⁸O-labeled H₂O and the regular H₂O with the isotopic composition in natural abundance were measured using a classical bromine pentafluoride fluorination technique (99). A small portion of the quartz sample was reacted with bromine pentafluoride to yield O₂, then the liberated oxygen was converted to carbon dioxide for mass spectrometric analysis. The isotopic ratios (¹⁸O/¹⁶O) are reported in the conventional δ¹⁸O-notation as permil deviation from the ratio of Standard Mean Ocean Water (SMOW).

The ¹⁸O-containing species on the abraded quartz (QGN) after being immersed in H₂¹⁸O and dried in the N₂ stream (ca. 25°C) was confirmed by the thermogravimetry-differential scanning calorimetry-quadrupole mass spectrometry-isotope ratio mass spectrometry (TG-DSC-QMS-IRMS) measurements. The STA449 F3 (Netzsh) and STA449 F3 (Pfeiffer) instruments were used for thermal analysis and mass spectrometric analysis, respectively. The temperature range for thermal analysis is 40 to 900°C, and the heating rate is 10°C min^–1 in 40 to 150°C or 20°C min^–1 in 150 to 900°C, respectively; the test atmosphere is N₂ gas; the flow rate is 20 mL min^–1. The mass in mass spectrometric analysis ranges from 1 to 400 amu with 64 channels.

Specific surface area measurements of the abraded quartz samples (QGN and QGO) were performed using BET technique with N₂ adsorbate gas. Nitrogen adsorption–desorption were measured at 77 K using an ASAP 2020 Surface Area & Pore Size Analyzer (Micromeritics Instrument Corporation). Samples were degassed in a vacuum at 200°C for 12 h prior to the measurement.

Measurements of Reactive Oxygen Species (•OH and H₂O₂).

Hydroxyl radical (•OH) production was measured by determining the concentrations of the oxidative product (p-HBA) of BA (100). Hydrogen peroxide (H₂O₂) concentrations were determined using a modified version of Leuco Crystal Violet (LCV, spectrographic grade, John Long) technique optimized for UV–Vis measurements (101). The detailed ROS detection procedures followed those described in He et al. (25).

Experiments Procedures of ¹⁸O Labeling Experiments.

The ¹⁸O-labeled water (H₂¹⁸O, at. 97% ¹⁸O), the •OH trapping agent (benzoic acid, BA), and its corresponding oxidation product (p-hydroxybenzoic acid, p-HBA) were obtained from Aladdin. The mother solution of BA (25 mM) was prepared by dissolving BA in H₂¹⁸O, while the catalyst for H₂O₂ dissociation, FeCl₂(H₂¹⁸O) (13.33 mM), was prepared by dissolving FeCl₂ in H₂¹⁸O. All samples for the quantification of the oxygen isotopic composition were prepared in the glove box (N₂ > 99.999%, H₂O < 0.1 ppm, O₂ < 0.1 ppm).

To determine the source of oxygen in ROS produced by the reactions between water and abraded quartz, ¹⁸O-labeled water (H₂¹⁸O) was used as the reactant. When the •¹⁸OH is formed in the H₂¹⁸O splitting, it can oxidize BA at a fast rate (•OH + BA → p-HBA, k = 4.3 × 10⁹ M^–1 s^–1), and the resulted p-HBA with a hydroxyl group (–¹⁸OH) formed at the para-position of the benzene ring could be identified by mass spectrometer (65). It is noteworthy that the quartz used in this study consists almost of Si¹⁶O₂ (the atomic percent of ¹⁶O in O is about ~99.75%), so the H₂¹⁸O is the major source of ¹⁸O in the reactions. The detail experimental procedures could be found in the SI Appendix (The experimental procedures of the ¹⁸O-labeling experiments: Set I–IV).

Measurements of the Oxygen Isotopic Composition in •OH and H₂O₂.

Based on the p-HBA approach that determines •OH concentrations, the methods for the measurements of the oxygen isotopic composition in •OH and H₂O₂ were developed. For the oxygen isotopic composition in •OH that is produced in the reactions between SBRs and water, it can be measured by quantifying the contents of p-HBA with –¹⁸OH (named as p-HBA(¹⁸OH)) and p-HBA with –¹⁶OH (named as p-HBA(¹⁶OH)) that are produced in the oxidation of BA by •OH. The oxygen exchange between the –COOH group of BA and ¹⁸O-labeled water, which may lead to potential isobaric interference of p-HBA(¹⁸OH), is properly monitored in our study (SI Appendix, Text and Fig. S4). The background signal that may overestimate p-HBA(¹⁸OH) quantification is only 3.6%, within experimental uncertainties shown in Fig. 3 and SI Appendix, Fig. S3, and therefore would not influence our interpretation and conclusion. For the oxygen isotopic composition in H₂O₂, it can be measured by quantifying the content of p-HBA(¹⁸OH) and p-HBA(¹⁶OH) that is produced in the oxidation of BA by •OH that is generated in the dissociation of H₂O₂ after adding FeCl₂.

The quantification was conducted on an Agilent 1200 series liquid chromatography (LC) and an Agilent 6410 triple quadrupole mass spectrometer (MS/MS). The mobile phase was acetonitrile (35%) and 0.1% dilute ammonia water (65%) throughout the process for 3 min. The flow rate was 0.3 mL min^–1 with an injection volume of 2 μL. The p-HBA was quantified under multiple reaction monitoring (MRM) conditions. The electrospray ionization (ESI) with an electrospray negative ion mode was used to produce gas-phase ions. The optimized parameters were as follows: gas temperature, 350°C; capillary voltage, –4,000V; nebulizer, 40 psi; gas flow,10 L min^–1; dwell time, 200 ms; fragmentor, 60 eV; the collision energy, –10 eV. The m/z transitions for qualifiers were 137/93 and 139/95 for the general p-HBA and the ¹⁸O-labeled p-HBA, respectively.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix. The experimental data has been deposited in Mendeley Data (https://doi.org/10.17632/fy6w4txbwf.1).