Decoupling of high-pressure H2 production from serpentinization and magnetite in subduction zones

doi:10.21203/rs.3.rs-4973494/v1

Download PDF

Article

Decoupling of high-pressure H₂ production from serpentinization and magnetite in subduction zones

https://doi.org/10.21203/rs.3.rs-4973494/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Serpentinization plays a central role in geological, geochemical, and microbiological processes at various depths and conditions. While mineralogical and geochemical patterns of serpentinization are known at low-pressure and temperature, equivalent processes taking place at much greater depths and elevated temperatures in subduction zones are less constrained. Here we present the results of thermodynamic calculations simulating chemically complex aqueous alteration of ultramafic rocks relevant to three evolutionary stages of subduction, from infancy to maturity, and for three different fluid sources, namely metabasite, metasediment, and serpentinite. We found that magnetite production and H₂ and CH₄ concentrations are decoupled from serpentinization at these conditions, and strongly dependent upon minimal peridotite compositional variations. This has major implications on the interpretation of geophysical investigations of hydrated mantle wedge domains. Carbon isotopes exhibit large variations (> 10‰) for minimal changing compositions in some cases, with important implications on the isotopic diversity of high-temperature abiotic CH₄.

Earth and environmental sciences/Solid Earth sciences/Geochemistry

Earth and environmental sciences/Solid Earth sciences/Geodynamics

Biological sciences/Microbiology/Biogeochemistry/Carbon cycle

Earth and environmental sciences/Solid Earth sciences/Geophysics

Earth and environmental sciences/Planetary science/Petrology

Serpentinization, the process of aqueous alteration of ultramafic rocks, is known to profoundly affect the rheology of mantle rocks, global geochemical cycles, and to produce fluids enriched in H₂ and abiotic hydrocarbons, from shallow sub-seafloor ^1–7 to high-pressure conditions in subduction zones ^8–10. A large body of literature has dealt with the mineralogical, geophysical, and geochemical patterns of serpentinization at shallow conditions, as well as on the critical parameters promoting H₂ and abiotic hydrocarbon production at conditions that largely overlap habitable conditions ^11–14. The possibility of serpentinization to take place at greater depths is also known and has been the subject of several geophysical investigations in subduction zones. For example, low shear-wave velocities and gravity and magnetic anomalies at present day subduction zones have been interpreted to testify for the presence of hydrous silicate and magnetite-rich rocks characteristic of serpentinization in the mantle wedge above subducting slabs ^15–17. However, the petrological, geophysical, and geochemical fingerprints of these deep processes are still largely unconstrained. In particular, the production of energy sources such as H₂ and abiotic hydrocarbons through serpentinization at depths outside the parameter space for life, as well as their purely abiotic isotopic fingerprints, remain little investigated.

Here we investigate the evolution, at different stages of subduction maturity, of the patterns of serpentinization, the resulting mineralogy and production of H₂-CH₄-rich fluids, and the carbon stable isotope signatures of serpentinization-related CH₄ at high pressure-temperature (P-T) conditions relevant to subduction zones from 300 to 700°C and from 1 to 3 GPa. We present the results of 206 640 thermodynamic models of fluid-rock interactions including dissolved ionic species to reveal the importance of key variables such as P-T conditions, fluid composition and amount, and initial mantle mineralogy. To explore the isotopic dependence of the produced abiotic CH₄ on the above parameters, we performed isotopic mass balances thanks to a new 𝛿¹³C equilibrium fractionation database obtained from ab-initio simulations and comprising the 12 most abundant carbon-bearing fluid species present in our models, namely CO₂, CH₄, CO, H₂CO₃, CaCO₃, CO₃²⁻, HCOO⁻, HCO₃⁻, CaHCO₃⁺, CaHCOO⁺, NaHCO₃, NaCO₃⁻, FeHCO₃⁺.

Our results highlight characteristic patterns of high P-T serpentinization relative to shallower equivalents and provide important insights on both the suitable conditions for mantle wedge serpentinization and their geophysical implications. Additionally, our work improves our understanding of the processes capable of producing deep energy sources such as natural H₂ and abiotic CH₄ and their characteristic isotopic fingerprints.

To model fluid-peridotite interactions at subduction zone conditions, fluids equilibrated with three different sources were considered, namely serpentinite, metabasite, and metasediment. Representative bulk-rock major element compositions for these lithologies were selected from ¹⁸ and comprise water-saturated depleted mid-ocean ridge mantle (DMM, for serpentinite), mid-ocean ridge basalt (MORB, for metabasite), Mariana Clay (for metasediment). These compositions were used to construct pseudosections using PerpleX 7.0.7 ¹⁹ and to derive equilibrium metamorphic assemblages and fO₂ for three sets of P-T conditions representative of thermal profiles for early (5.6 Ma), moderate (11.8 Ma) and mature (32.1 Ma) subductions ²⁰. The obtained assemblages and fO₂ were used to model equilibrium aqueous fluid component speciation for each fluid source and P-T couple using the software EQ3 ²¹ and the Deep-Earth Water (DEW) model ²². These fluids were then reacted with 861 different ultramafic assemblages spanning the entire olivine-orthopyroxene-clinopyroxene (Ol-Opx-Cpx) ternary diagram (+ 2 vol% spinel) and for fluid/rock ratios (in mass) from 1 to 10, using the EQ6 software ²¹.

From early to mature subduction, and from lower to higher P-T conditions, calculated log fO₂ values relative to the fayalite-quartz-magnetite buffer (𝛥FMQ) range between − 3.8 and 3.8 for serpentinite, -0.5 to 2.6 for metabasite, and − 0.4 to 3.5 for metasediment. The solute load of fluid in equilibrium with the selected lithologies increases with increasing P and T. In general, fluids from serpentinite are richer in dissolved Mg than metabasite (5-250 times more depending on P-T) and metasediment (15-7000 times more). Fluids in equilibrium with metabasite and metasediment have very similar dissolved Si and Al, both present in higher concentrations relative to fluid in equilibrium with serpentinites below 500°C, and in lower concentrations at higher T. Fluids equilibrated with metabasite are richer in dissolved Ca compared to both serpentinites (4–83 times more) and metasediments (1–47 times more).

Reacting these fluids with spinel-bearing ultramafic rocks reveals strong heterogeneity within the Ol-Opx-Cpx space (Figs. 1 and 2). The degree of serpentinization, here defined as the modal proportion of forming antigorite, is strongly dependent upon the fluid source, with fluids from serpentinite producing more antigorite than metabasite and metasediment-derived fluids for all P-T and fluid-rock ratio conditions (Fig. 2). Antigorite production is maximized for Ol-Opx mixtures. Nevertheless, Ol-Opx proportions for maximized antigorite productions vary depending on the fluid source and P-T conditions. Fluids from metabasite and metasediment tend to produce more antigorite for rocks with higher olivine proportions compared to fluids from serpentinite, which show the highest antigorite production for equal Ol-Opx proportions. This trend is enhanced for higher fluid/rock ratios for fluids from serpentinites, especially at high P-T conditions in the stability field of olivine and antigorite where there is barely no antigorite production for olivine-rich rocks (Fig. 2d), but where other hydrous phases form instead (Fig. S1). Figures 2g-i show a natural example of metamorphic serpentinization of subducted peridotite from the Belvidere Mountain complex, Northern Vermont ²³. Alternating harzburgite and dunite layers show strikingly different serpentinization degrees, with Opx-bearing (harzburgite) layers forming abundant antigorite and Opx-free (dunite) layers being rather unaffected by serpentinization. The conditions of serpentinization were estimated at about 450°C and 1 GPa ²³.

In the stability field of olivine and antigorite, the degree of serpentinization decreases with increasing fluid-rock ratios for fluids from metabasite (Fig. 2e) and metasediment (Fig. 2f), with no serpentinization for fluid/rock ratios above 7. This was observed for every peridotite composition and is linked with more abundant talc formation instead. Primary olivine consumption at reaction completion increases with fluid/rock ratios and is completely reacted at fluid/rock ratios higher than 5 at low P-T conditions within the antigorite + brucite stability field. In contrast, at higher P-T conditions in the stability field of olivine + antigorite, primary olivine is never completely consumed for olivine-rich peridotite compositions, even at high fluid/rock ratios (see figures S1, S4 and S7). At these conditions, a large amount of new metamorphic olivine is formed with modal proportions depending upon the starting peridotite composition (Fig. 1h).

Magnetite contents are a clear function of P-T conditions (Fig. 3), with higher magnetite production at lower P-T conditions (maximum of 1.1 to 1.4 vol% at 300°C and 1 GPa, Fig. 1b) relative to high P-T conditions (maximum of 0.004 to 0.14 vol% at 650°C and 2 GPa, Fig. 2b). Magnetite production is highest for serpentinite-derived fluids, about 5–540% more relative to metabasite and metasediment-derived fluids, respectively. In the stability field of olivine + antigorite (above 400–500°C at the investigated pressures), magnetite modes increase with increasing fluid rock-ratios at every P-T condition, even with no coexisting antigorite. At these conditions, magnetite modes are not correlated with serpentinization. Instead, magnetite formation is inversely correlated with the formation of metamorphic olivine, i.e., where metamorphic olivine contents are high, magnetite contents are low. The high magnetite production is observed for Ol-Cpx mixtures. At these P-T conditions, metamorphic olivine is significantly enriched in fayalite compared to the initial peridotite composition. Similarly to antigorite, metamorphic olivine is higher for serpentinite-derived fluids compared to metabasite- and metasediment-derived fluids at every P-T condition (Fig. 1h). Metamorphic Cpx formation is maximized for Cpx-rich reacting assemblages. Formation of hydrous phases other than antigorite is dependent on the starting ultramafic mineralogy, with higher talc content close to the Opx corner and higher tremolite for Opx-Cpx mixtures (Fig. S1). Both talc and tremolite formation is limited for serpentinite-derived fluids (Fig. S1). Instead, they are more abundant and have similar proportions for metabasite- and metasediment-derived fluids, with talc being present at all P-T conditions and more abundant at medium P-T conditions (Fig. S2 and S3) and tremolite present only at low to medium P-T conditions (Fig. S2 and S3). Similarly to antigorite and magnetite, tremolite and talc contents are strongly dependent upon the fluid/rock ratios. Higher fluid/rock ratios stabilize more talc and tremolite and for a broader range of reacting ultramafic rocks for metabasite- and metasediment-derived fluids. Similarly, brucite formation can be abundant for reacting dunite compositions (up to 6 vol%) and serpentinite-derived fluids, which contain high amounts of dissolved Mg (Fig. 1d). Chlorite contents exhibit small variations inside each ternary, most likely due to the fact that it is the primary Al-bearing phase produced by the dissolution of the primary spinel. The slightly higher than expected chlorite contents are most likely due to the fact that Al is not incorporated into antigorite in the DEW thermodynamic models.

Figure 1. Compositions at the end of the -fluid-mantle interaction at 1 GPa and 300°C (a, b, c, d) and 1 GPa and 500°C (e, f, g, h) and for the three considered fluid sources; a) antigorite modes (vol%); b) magnetite modes (vol%); c) H₂ content in the reacted fluid in molal (mol/kg of solvent); brucite modes (vol%); e) antigorite modes (vol%); f) magnetite modes (vol%); g) H₂ and h) metamorphic olivine (M-Ol) modes (vol%). The decoupling between antigorite contents and H₂ production is marked at 1 GPa and 500°C. Note that magnetite and H₂ contents are higher where metamorphic olivine content is lower. The white fields on each ternary diagram is the compositional range for natural mantle rocks ²⁴.

Upon completion of fluid-rock reactions, low fO₂ and high H₂ and CH₄ concentrations in the fluid are primarily correlated with magnetite production and with the fO₂ of the reactant fluid. The patterns of magnetite production, fO₂, and H₂ and CH₄ concentrations vary substantially upon the considered fluid source, which makes comparisons challenging. Nevertheless, serpentinite-derived fluids lead to lower fO₂ and maximize magnetite and H₂ and CH₄ production at almost all P-T conditions investigated in this study. H₂ contents follow similar trends for serpentinite-derived fluids compared to metabasite- and metasediment-derived fluids (Fig. S1).

CH₄ and CO₂ are the most abundant carbon species in the fluid at low P-T conditions, with CH₄ and CO₂ being dominant at low fO₂ and high fO₂, respectively. At higher temperatures around 500°C and 1 GPa, H₂CO₃, for fluids from metasediment and serpentinite become closer to CO₂, concentrations at high fO₂. At the same conditions, NaHCO₃ concentrations are the most abundant C species for high fO₂ and fluids from serpentinite. At even higher P-T conditions above 600°C and 2 GPa, HCOO⁻ concentrations become closer to CH₄ concentrations at low fO₂. This is also a function of the fluid/rock ratio, with CH₄ and CO₂ decreasing with increasing fluid/rock ratio. In addition to fO₂, HCOO⁻, H₂CO₃, HCO₃⁻, NaHCO₃, CO₂ CH₄ and H₂ contents are also dependent on the pH of the reacted fluid. Dissolved Fe in the fluid positively correlates with magnetite content variations at low P-T conditions while at higher P-T conditions it correlates with metamorphic olivine contents.

Figure 2. Antigorite contents for all fluid-rock ratios and all three fluid sources for 1GPa and 300°C (a) and 500°C (b) for rock compositions with 0% Cpx and variable amount of Ol and Opx here shown as Orthopyroxene %. The shift of the maximum amount of antigorite visible on (b) for fluids from serpentinites is nicely visible. Natural example of metamorphic serpentinization from the Belvidere Mountain complex, Nothern Vermont (g-i). Primary compositional variations in alternating harzburgite and dunite layers result in contrasting serpentinization patterns, with strongly serpentinized harzburgite layers and weakly serpentinized dunite layers. Photomicrographs are in plane-polarized light (h) and cross-polarized light (i).

In contrast to what is usually assumed, modeling results at the considered high P-T conditions highlight a remarkable mismatch between antigorite formation, i.e., serpentinization, and the patterns of magnetite formation, fO₂, and H₂ and CH₄ concentrations in the reacted fluid within the Ol-Opx-Cpx space. This mismatch is especially marked at higher P-T conditions, especially in the stability field of olivine + antigorite. At these P-T conditions, magnetite proportions are very small, below 0.2 vol% for serpentinite-derived fluids, and therefore H₂ production is also very small. Magnetite production (and therefore H₂ and CH₄) increases with increasing fluid rock-ratios at every P-T conditions.

As for minerals and fluid species, carbon isotopic signatures (𝛿¹³C) of fluid species vary as a function of the reacting ultramafic mineral assemblage (Fig. 4). The maximum range of 𝛿¹³C within a single Ol-Opx-Cpx diagram, and for a fixed 𝛿¹³C for the infiltrating fluid, is about 15‰ at low P-T conditions (Fig. 4c). This shift affects all carbon bearing fluid species so that only the absolute 𝛿¹³C of each species changes. Carbon isotopic variations for individual species inside the ternary follow but do not completely match the fO₂ variations. In fact, pH variations also control isotopic variability. At higher P-T conditions, since the different carbon-bearing aqueous species have similar concentrations, 𝛿¹³C variations are very small. The changes in isotopic signature can be very abrupt, for example, for a serpentinite-derived fluid at 300°C and 1 GPa, the 𝛿¹³C signature of CH₄ after reaction with a dunite is 5‰ higher compared to a harzburgite with only 10% Opx. Therefore, small changes in peridotite compositions can have drastic effects on the isotopic signatures of carbon species.

When the cold mantle wedge does not like to hydrate. The formation of hydrous phases during the modeled fluid-mantle rock interactions is strongly dependent upon the fluid source, even for equivalent initial rock compositions within the Ol-Opx-Cpx diagram. Serpentinite-derived fluids are found to maximize antigorite production, especially at high fluid/rock ratios, whatever P-T conditions in the stability field of antigorite. Fluids from metabasite and metasediment show the same trend at low P-T conditions, whereas they result in opposite reaction patterns at higher P-T conditions (low antigorite for higher fluid/rock ratios). Rocks falling across the harzburgite-olivine orthopyroxenite compositional range, roughly around 1:1 for Ol-Opx proportions, produce more antigorite than typical mantle harzburgite or lherzolite (Fig. 1). This feature differs compared to low P-T serpentinization patterns and results from the higher silica content of antigorite relative to lower temperature serpentine polysomes such as lizardite. Favorable Mg/Si ratios for antigorite formation may also be promoted by Mg loss during serpentinization of relatively Opx-poor harzburgites ⁴. In addition, at P-T conditions within the olivine + antigorite stability field, expected to represent low P during early subduction or higher P with increasing subduction maturity, antigorite is in competition with metamorphic olivine for Mg. At these conditions (antigorite + olivine), higher amounts of fluids deriving from a serpentinite source will produce less antigorite for typical abyssal and orogenic peridotite compositions (Fig. 1d). Thereby, in contrast to the conventional assumption that serpentinization is enhanced for olivine-rich compositions, which remains valid at low-T conditions ^25,26, high-T-P serpentinization at mantle wedge conditions is instead favored for Opx-rich peridotites ²⁷, especially during early subduction since it is usually warmer (0–10 Ma from subduction initiation). Talc formation and proportions are maximized for Opx-rich rocks and metasediment-derived fluids, which favor high SiO₂ activity ²⁸. Finally, tremolite contents are maximized for mixed Cpx-Opx compositions and metabasite-derived fluids. Although tremolite requires Ca from Cpx, tremolite and Cpx compete for calcium over a wide pressure-temperature field, so that tremolite contents are never high for Cpx-rich compositions. Similarly, magnetite and metamorphic olivine compete for Fe. This explains why magnetite proportions are strongly affected by the amount of metamorphic olivine at medium to high P-T conditions, and why the Mg# of metamorphic olivine predicted by our calculations varies accordingly. This clearly shows that serpentinization, i.e., the formation of serpentine minerals, is controlled by variations in the chemistry of the serpentinizing fluid by several 10s of volume percent and that the competition with other phases for redox sensitive and non-redox sensitive elements ²⁹ directly influences the degree of serpentinization. Simple modeling relating water availability (e.g, H₂O saturation) in mantle rocks to serpentinization ^9,20,30 can significantly under or overestimate the amount of serpentine being produced during the aqueous alteration of mantle rocks at subduction zone conditions. This is especially true for low P-T conditions where fluids released from subducted sediments or from subducted oceanic crust will tend to produce more talc or tremolite than serpentine. The predicted magnetite contents are extremely variable within the olivine-orthopyroxene-clinopyroxene space. Our data show that there is no need for kinetic explanations for the absence of magnetite in certain serpentinized peridotite as previously suggested ³¹. Instead, the absence or presence of magnetite at high P-T conditions can be explained by the Mg/Si ratio of the fluid-rock system, which controls the presence or absence of metamorphic olivine as well as the antigorite proportions. Iron uptake by metamorphic olivine precludes magnetite formation or significantly reduces its modal proportions. At elevated temperature (e.g., 500°C), antigorite-poor, metamorphic olivine-rich rocks forming from olivine-rich peridotite precursors may contain substantially higher magnetite proportions relative to antigorite-rich rocks forming from Opx-rich harzburgite. This feature may result in geophysical decoupling between seismic velocities and magnetic anomalies. Also, magnetite-bearing metamorphic peridotites may not necessarily reflect conditions beyond the serpentine stability field; they may instead result from peridotite-fluid compositions unconducive to serpentinization.

Figure 3. Magnetite (Mt) and antigorite (Atg) modes for each set of P-T conditions considered. The decoupling between magnetite and antigorite production is especially visible for panels a) and g) where magnetite modes are higher on the olivine-clinopyroxene joint while antigorite production is primarily between olivine and orthopyroxene, especially on panel a). At lower temperature and pressure(c-f), antigorite and magnetite modes match more closely.

H ₂ and CH₄ production in subduction zones. It is generally assumed that, besides very low-T conditions, H₂ and CH₄ production in response to aqueous alteration of ultramafic rocks follows serpentinization and magnetite formation. From our results, however, it emerges that H₂ and CH₄ production are not directly linked to antigorite formation or abundance. This is because of the anticipated decoupling between antigorite and magnetite at high P-T conditions. In particular, if at low P-T conditions H₂ and CH₄ production match with serpentinization patterns (see Fig. S1 for CH₄), at higher P-T conditions in the stability field of olivine + antigorite, a clear decoupling is observed. At these conditions, H₂ and CH₄ are correlated with fO₂, which is linked to the amount of magnetite and inversely correlated with metamorphic olivine contents (Fig. 1). Production of H₂ is higher at lower P-T conditions, i.e., below 400°C, and higher for higher fluid/rock ratios, consistent with previous studies that did not expand to higher T and especially higher P ^11,29. In fact, ¹¹ showed that the stability of olivine significantly decreases the amount of H₂ produced, but this was only inferred to be T dependent. Here we show that the starting peridotite will play a major role in the olivine-magnetite patterns and therefore H₂ production. The fact that young subduction zones are warmer compared to mature subduction zones ²⁰ implies that more H₂ may be produced from cold, mature subduction zones, even though the amount of serpentinization is smaller due to lower production of aqueous fluids from the slab.

Molecular hydrogen and CH₄ production through subduction zone serpentinization may represent a substantial fraction of global fluxes of these reduced molecules ^10,32. Their production relative to more oxidized fluid species may impact global volatile cycling. These molecules may also be key players for energy metabolism in the deep subsurface biosphere at convergent margins ^33,34. The observed decoupling between serpentinization and magnetite production may provide important insight on the interpretation of geophysical data for the identification of potential sources of these deep energy sources. In particular, heavily serpentinized domains of the mantle wedge may not necessarily represent major contributors of H₂ and CH₄ relative to drier domains within the serpentine stability field. Such interpretations should take into account subduction maturity, fluid sources, and mantle mineralogy.

Carbon isotopic signatures. Carbon isotopes are an important parameter in the interpretation of the origin of CH₄ through biotic or abiotic processes and biotic or abiotic carbon sources ³⁵. The possibility for a process to generate high CH₄ isotopic diversity, and especially from a constant carbon isotope source, may provide important insights on the range of biotic/abiotic methane isotope signatures. Our results show that substantial CH₄ carbon isotopic variation can be generated for the same fluid source within the Ol-Opx-Cpx space. The final isotopic diversity of CH₄ is controlled by the isotopic mass balance of the system, which varies across the compositional space for a constant fluid source. The most important factors controlling the carbon fluid speciation and isotopic signatures are fO₂ and pH. The first one dictates the amount of oxidized relative to reduced species, while the second controls the concentration of species other than CO, CO₂, and CH₄. Nevertheless, the role of fO₂ at determining large variations of oxidized relative to reduced species also depends on the absolute fO₂ value. This feature is clearly visible on Fig. 4. For serpentinite-derived fluids, at 1 GPa and 300°C the largest 𝛿¹³C_CH4 variations is less than 0.2‰ (Fig. 4a) and fO₂ ranges from − 3.8 to -6.2 𝛥FMQ. Instead, for the same serpentinizing fluid, at 1 GPa and 500°C the largest 𝛿¹³C_CH4 variation is ~ 5‰ (Fig. 4b), significantly larger than 300°C, and fO₂ ranges only from − 1.3 to -2.1 𝛥FMQ. In the first case, the infiltrating fluid is initially too reduced to produce large variations in the concentration of carbon species, and CH₄ is the dominant carbon species for all peridotite compositions. The absolute fO₂ explains the small isotopic variations of Fig. 4a compared to the large variations of Fig. 4b. Therefore, small fO₂ variations for initial fluids containing similar proportions of reduced and oxidized fluids (e.g., similar proportions of CH₄ and CO₂ for example) will produce larger isotopic variations relative to large fO₂ variations from initial strongly reduced or oxidized fluids. Our results show that the carbon isotopic signatures are not only a function of the initial fluid composition but may vary substantially as a function of the initial redox state and reactant ultramafic rock. For example, a metasedimentary fluid with an initial bulk 𝛿¹³C of 0‰ at 1GPa and 300°C and reacting with a lherzolite, harzburgite, or dunite would result in a 𝛿¹³C_CH4 of -1‰, very similar to the bulk infiltrating fluid. This signature would point to an inorganic carbon source for abiotic CH₄. However, at the same conditions, the same fluid reacting with a pyroxenite would lead to 𝛿¹³C_CH4 of -15‰, which may erroneously suggest the contribution of an organic carbon source. As seen on Fig. 4b, large 𝛿¹³C can reflect even smaller mineralogical changes in reactant ultramafic rock. As also discussed for serpentinization, variations of a few vol.% of Opx in the reactant rock can produce several ‰ differences in the final 𝛿¹³C of aqueous species. Since isotopic fractionation is a strong function of temperature with higher fractionation at lower temperature, cold, mature, subduction zones will tend to produce larger isotopic variations in the mantle wedge than warm, immature, subduction zones.

Figure 4. Example of CH₄ 𝛿¹³C, log fO₂, and pH after fluid-mantle interaction.At 1 GPa and 300°C (a) and at 1 GPa and 500°C (b) for a fluid from serpentinite and at 1 GPa and 300°C for a fluid from metasediment (c). All models are computed for an initial fluid with a 𝛿¹³C = 0‰. Variations inside the ternaries can be extremely limited (a), important, i.e., 5‰ in (b) or extremely high with total variations as high as 15‰ in (c).

Geophysical and biological implications. The modeled evolutions of high P-T serpentinization are substantially different from low P-T equivalents and are supported by natural data from subduction zone serpentinized peridotites (Fig. 2). Serpentinization, magnetite, and H₂ formation are central in a multitude of geobiological processes spanning rheology of the lithosphere, geochemical cycling, and the genesis of energy sources playing an important role in biology. Fluid-mantle interactions above subducting plates represent the deepest environments where serpentinization may occur, and understanding these interactions has implications for studying mantle domains along more than 55,000 km of present-day active margins. Serpentinization of the mantle wedge above subducting slabs has been the subject of multiple studies^9,15–17,36 with contrasting results. Some authors have proposed that mantle wedge serpentinization may be widespread ³⁶, whereas others point to minimal serpentinization except for hot subduction zones where substantial amounts of water can be flushed from dehydrating slabs at shallow depths ⁹. Our results indicate that the amount of water released from the slab is not necessarily the key factor. Instead, the strong influence of fluid chemistry and mantle mineralogy can lead to counterintuitive patterns, such as limited serpentinization within olivine-dominated assemblages, or sharp variations in serpentinization from minimal variations in Opx-Ol ratios.

Geophysical signals from present-day forearc mantle wedges have been used to assess the degree of serpentinization at convergent margins. Their interpretation, however, currently relies on extrapolation of constraints on serpentinization patterns at lower P-T conditions than expected in subduction settings. For example, low shear-wave velocities and gravity of the forearc mantle wedge at present-day subduction zones have been interpreted as the result of serpentinization ^15,36. Based on our findings however, the distribution of such low V_p/V_s regions relative to drier mantle assemblages may not necessarily reflect preferential fluid availability and circulation across the mantle wedge. For example, at 1 GPa (about 35 km depth) and 500°C –early subduction–, dunite may be unconducive for serpentinization while very small amounts (+ 10 vol.%) of additional Opx result in about 20–25 vol.% antigorite production. Such a variation may profoundly alter the V_p/V_s signal of mantle wedge domains being flushed by comparable fluid fluxes. Magnetite formation through serpentinization has also been used to interpret magnetic anomalies at subduction zones ^16,17. Magnetite vs brucite formation across a 200°C threshold is a well-established concept at low P-T conditions ¹². This concept has been used to higher P-T conditions ¹⁶ combined with satellite magnetic anomalies to infer the degree of serpentinization (assumed to correlate with magnetite content) within mantle wedge domains. However, our results suggest serpentinization and magnetite production may be decoupled at high P-T conditions. Furthermore, potential mismatches between seismic velocities and magnetic anomalies may reflect subtle fluid source and/or mantle mineralogy variations rather than presence or absence of aqueous fluids and serpentinization.

This decoupling also affects the production of deep H₂ and related abiotic CH₄. High P-T, heavily serpentinized domains may produce much smaller amounts of such reduced fluids compared to drier domains affected by fluid percolation but chemically unconducive for serpentinization. This feature may impact the identification of potential source regions of deep energy for deep microbial life. The chemical diversity of high P-T serpentinization, and its thermal evolution from subduction initiation to maturity, also impacts the identification of this deep, abiotic energy relative to biogenic sources within the biosphere, for example through carbon isotope fingerprints. Our isotope modeling results indicate that substantial isotopic variation in 𝛿¹³C_CH4 (up to more than 15‰) can arise from simple mantle mineralogy variations within the Ol-Opx-Cpx space at constant fluid composition, P, and T. This feature bears important implications on the interpretation of biotic relative to abiotic processes in the subsurface, and potentially on the interpretation of methane isotopic fingerprints on other celestial bodies.

Thermodynamic modeling. To obtain the stable mineral assemblages and the corresponding $\:{f}_{{O}_{2}}$, pseudosections were computed for the three fluid source lithologies using the open-source software Perple_X, version 7.0.7 ¹⁹. The Holland and Powell thermodynamic database ³⁷, version 633, was used (hp633ver.dat) to compute the 3 different pseudosections. The system of components for the serpentinite was NCFMASHO, for the metasediments computation it was MnNCKFMASHTO and for the oceanic crust it was NCKFMASHTO. The bulk compositions are shown in Extended Data Table 4. The equilibrium volume composition for the serpentinites, metamafics and metasediments (Mariana clay) was taken from ¹⁸. Excess oxygen was added for each composition. For the ultramafic, we used Fe³⁺/Fe^tot = 3.5 % as t is the minimum from ³⁸. A Fe³⁺/Fe^tot = 40 % was ued for the metasediments as it is the proposed value from ³⁹. Finally, for the metamafics we used a value of Fe³⁺/Fe^tot = 14 % based o the range of ratios from ⁴⁰. $\:{f}_{{O}_{2}}$ was computed using the chemical potential of O₂ at selected P-T conditions using the geothermal gradients from ²⁰. Three P-T points were selected from the three different geothermal gradients at 5.6, 11.8 and 32.1 Ma after subduction initiation, respectively. $\:{f}_{{O}_{2}}$ was computed as follow:

$$\:{f}_{{O}_{2}}=exp\left(\frac{{(\mu\:}_{{O}_{2}}^{P,T}-{\mu\:}_{{O}_{2}}^{1,T})}{RT}\right)$$

( 1)

Where $\:{\mu\:}_{{O}_{2}}^{1,T}$ represents the reference chemical potential of O2 at 1 bar and corresponding temperature, $\:{\mu\:}_{{O}_{2}}^{P,T}$ is the chemical potential of O₂ at the P and T of interest and R is the gas constant.

To model the fluid infiltration, the suite of software from EQ3/6 was used. EQ3 computes the speciation of the ionic fluid from a stable assemblage while EQ6 uses the fluid from EQ3 to model the interaction of this fluid with a rock. The mineral assemblages and mineral compositions for EQ3 were computed using the pseudosections from Perple_X at the specific P-T conditions. The chlorinity and the total dissolved carbon in the fluid were set to 0.01 and 0.001 molal (mol/kg of water), respectively. The pickup files computed from EQ3 were then used to build the input files for EQ6.

Carbon isotopic composition for each fluid species were computed using the fractionation factors from first principles calculations (see next section) and molalities of each species in the fluid. Since the total carbon dissolved in the fluid was small, no carbonate or graphite was produced during the computations. Therefore, the δ¹³C of each species was only a function of the molalities and the input bulk fluid δ¹³C, in this case δ¹³C_fluid = 0‰. It was derived from the mass action law and computed at each reaction progress of the fluid-rock interaction as follows:

$$\:{{\delta\:}^{13}C}_{ref}={{\delta\:}^{13}C}_{fluid}+{\sum\:}_{i\ne\:ref}^{n}{X}_{i}^{C}{\varDelta\:}_{ref-i}$$

( 2)

$$\:{{\delta\:}^{13}C}_{i}={{\delta\:}^{13}C}_{ref}-{\varDelta\:}_{ref-i}$$

( 3)

Where, δ¹³C_ref is the carbon isotopic composition of CO₂ or CH₄, $\:{X}_{i}^{C}$represents the fraction of carbon in species i relative to the total amount of carbon in the fluid, δ¹³C_i is the carbon isotopic signature of species i and $\:{\varDelta\:}_{ref-i}$ is the equilibrium fractionation of CO₂ or CH₄ with species i. In this case, all species in the fluid have only one carbon, so that the fraction of carbon can be computed using the following equation:

$$\:{X}_{i}^{C}=\frac{{m}_{i}}{{\sum\:}_{i=1}^{n}{m}_{i}}$$

( 4)

Where $\:{m}_{i}$is the molality of species i in the fluid. From Eqs. (2) and (3) it is clear that major changes in the concentration of the major species will create large isotopic variations for all species.

H ₂ production from Fe³⁺ in antigorite. Our computations do not consider Fe³⁺ in minerals other than iron oxides, and a pure magnesian endmember for antigorite. Based on the Mg# of metamorphic olivine and the known partitioning coefficient between olivine and antigorite and the Fe³⁺/Fe^tot of antigorite, we made a back of the envelope estimation of how much H₂ would be produced and compare this to our results. For example, at 1 GPa and 500°C, for a fluid-rock ratio of 1, when peridotite is infiltrated by serpentinites derived fluid, the maximum amount of antigorite is produced for a peridotite composition of 50% Opx and 50% olivine. This infiltration produces 40 vol% of antigorite (0.178 mol) and a metamorphic olivine with a Mg# of 0.7. An antigorite in equilibrium with such olivine composition would have a Mg# of 0.9 ⁴¹ and a Fe³⁺/Fe^tot of 0.4 ⁴². This would equal 0.007 mol/kg of H₂ produced from the oxidation of antigorite. Therefore, the H₂ produced from Fe³⁺ in antigorite is more than 5 times lower than the results obtained from our models for the same initial peridotite composition and one order of magnitude lower than the H₂ production for magnetite maximum along the Ol-Cpx joint (Fig. 1c). This shows that the decoupling between antigorite production and magnetite/H₂ generation would still be largely observed even when considering iron and iron valence in antigorite.

Calculation of equilibrium isotopic fractionation factors. The isotopic fractionation factor of carbon, $\:{\alpha\:}^{13}{C}_{a-b}$, at equilibrium between two phases $\:a$ and $\:b$ is defined as:

$$\:{\alpha\:}^{13}{C}_{a-b}=\frac{{\beta\:}^{13}{C}_{a}}{{\beta\:}^{13}{C}_{b}}$$

( 5)

where $\:{\beta\:}^{13}{C}_{a}$ and $\:{\beta\:}^{13}{C}_{b}$, also called $\:\beta\:$-factors, correspond to the reduced partition function ratios between phase $\:a$ or $\:b$ respectively, and a perfect gas of $\:C$ atoms.

For an isolated molecule, the reduced partition function ratio can be expressed, after using the Teller-Redlich product rule ⁴³, as:

$$\:{\beta\:}^{13}C={\left[\prod\:_{i=1}^{3N-6}\frac{{\nu\:}_{i}^{12}}{{\nu\:}_{i}^{13}}\times\:\frac{{e}^{-h{\nu\:}_{i}^{12}/\left(2{k}_{B}T\right)}}{1-{e}^{-h{\nu\:}_{i}^{12}/\left({k}_{B}T\right)}}\times\:\frac{1-{e}^{-h{\nu\:}_{i}^{13}/\left({k}_{B}T\right)}}{{e}^{-h{\nu\:}_{i}^{13}/\left(2{k}_{B}T\right)}}\right]}^{1/n}$$

( 6)

where $\:n$ is the number of exchangeable atoms in the molecule, $\:{\nu\:}_{i}^{12}$ and $\:{\nu\:}_{i}^{13}$ correspond to the harmonic frequency of the vibrational mode $\:i$ before and after substitution of the isotope ¹²C by the isotope ¹³C, respectively.

Carbon reduced partition function ratios were thus determined for the following carbon-bearing species: CO₂, CH₄, CO, H₂CO₃, CaCO₃, CO₃²⁻, HCOO⁻, HCO₃⁻, CaHCO₃⁺, CaHCOO⁺, NaHCO₃, NaCO₃⁻ and FeHCOO⁺. The equilibrium isotopic fractionation factors between these species could then be used as reference values for $\:{{\Delta\:}}_{ref-i}$ in Eq. (3), with $\:ref$ either CO₂ or CH₄, following the expression:

$$\:1000ln{\alpha\:}^{13}{C}_{ref-i}\approx\:{\delta\:}^{13}{C}_{ref}-{\delta\:}^{13}{C}_{i}={{\Delta\:}}_{ref-i}$$

( 7)

$\:\beta\:$ -factors were computed at the Density Functional Theory (DFT) level using the B3LYP hybrid functional ^44,45 in conjunction with the 6-311 + + G(2d, 2p) Pople-type basis set ^46,47, a scheme that has been shown to efficiently describe the structural and isotopic properties of dissolved inorganic carbon species ⁴⁸. Geometry optimizations were performed with the Gaussian 16 code ⁴⁹ using Tight optimization criteria and an Ultrafine integration grid mesh. For all species the spin multiplicity was singlet. Vibrational frequencies were then computed for each isotopomer and used to calculate the reduced partition function ratio as per Eq. (6). Implicit solvation was included via the Polarizable Continuum Model (PCM) with a dielectric constant equal to that of pure water. Regarding ion pairs, several starting configurations containing an explicit first solvation shell were explored, involving either monodentate or bidendate pairing between the carboxylic group and the cation, as well as total coordination numbers ranging from 4 to 6 around Na⁺ and from 6 to 8 around Ca²⁺. For Fe²⁺, the total coordination number remained 6. However, both the coordination number by water and the mono- or bidentate pairing, were found to lead to $\:{\beta\:}^{13}C$ variations of less than 3.5‰ at 25°C (see details in SI). Therefore, the $\:\beta\:$-factor of only one configuration per each ion pair species (shown in Figure S4) was arbitrarily selected for this application.

Although they were not included in the final thermodynamic modelling, geometry optimizations and vibrational frequency calculations were also conducted in gas phase for comparison with literature data and implicit solvent calculation. The subsequent $\:\beta\:$-factors are reported in SI.

Data availability

The data that supports the findings of this study have been deposited in a Mendeley Database accessible at 10.17632/pdwnfsm84m.1. It includes Python codes used to generate the data for the ternaries at each P-T conditions as well as all input necessary to reproduce the data and all outputs as csv files and individual plots can easily be produced using a Python file located in this repository. It also includes a step-by-step description of how to reproduce the study. Additional figures for all P-T conditions are also included in this repository.

Competing Interests

The authors have no competing interests.

Author Contributions

G.S. and A.V.B designed the study, G.S. performed the thermodynamic modeling, M.B. and J.A. designed and performed the ab initio calculations, G.S. drafted the manuscript, G.S., A.V.B., M.B., J.A. and S.W. reviewed and edited the manuscript.

Acknowledgments

This work is part of a project funded by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (Grant agreement No. 864045; project acronym DeepSeep). A MUR FARE grant (No. R20ZIYMPAR; acronym DRYNK), and a MUR PRIN2022 grant (No. 20224YR3AZ; acronym HYDECARB) to AVB are also acknowledged. Ab initio calculations were performed using the HPC resources from CALMIP (grant 2023 – P1037).

SW was supported by Australian Research Council (grant FT210100557).

Moody, J. B. Serpentinization: a review. Lithos 9, 125–138 (1976).
Charlou, J. L. et al. Intense CH4 plumes generated by serpentinization of ultramafic rocks at the intersection of the 15°20′N fracture zone and the Mid-Atlantic Ridge. Geochim. Cosmochim. Acta 62, 2323–2333 (1998).
McCollom, T. M. Laboratory Simulations of Abiotic Hydrocarbon Formation in Earth’s Deep Subsurface. Rev. Miner. Geochem. 75, 467–494 (2013).
Malvoisin, B. Mass transfer in the oceanic lithosphere: Serpentinization is not isochemical. Earth Planet. Sci. Lett. 430, 75–85 (2015).
Klein, F., Tarnas, J. D. & Bach, W. Abiotic Sources of Molecular Hydrogen on Earth. Elements 16, 19–24 (2020).
Tutolo, B. M., Seyfried, W. E. & Tosca, N. J. A seawater throttle on H2 production in Precambrian serpentinizing systems. Proc. Natl. Acad. Sci. 117, 14756–14763 (2020).
Leong, J. A. et al. H2 and CH4 outgassing rates in the Samail ophiolite, Oman: Implications for low-temperature, continental serpentinization rates. Geochim Cosmochim Ac 347, 1–15 (2023).
Deschamps, F., Godard, M., Guillot, S. & Hattori, K. Geochemistry of subduction zone serpentinites: A review. Lithos 178, 96–127 (2013).
Abers, G. A., Keken, P. E. van & Hacker, B. R. The cold and relatively dry nature of mantle forearcs in subduction zones. Nat Geosci 10, 333–337 (2017).
Brovarone, A. V. et al. Subduction hides high-pressure sources of energy that may feed the deep subsurface biosphere. Nat Commun 11, 3880 (2020).
McCollom, T. M. & Bach, W. Thermodynamic constraints on hydrogen generation during serpentinization of ultramafic rocks. Geochim Cosmochim Ac 73, 856–875 (2009).
Klein, F. et al. Magnetite in seafloor serpentinite—Some like it hot. Geology 42, 135–138 (2014).
Plümper, O. et al. Subduction zone forearc serpentinites as incubators for deep microbial life. Proc. Natl. Acad. Sci. 114, 4324–4329 (2017).
Sleep, N. H., Meibom, A., Fridriksson, Th., Coleman, R. G. & Bird, D. K. H2-rich fluids from serpentinization: Geochemical and biotic implications. Proc. Natl. Acad. Sci. 101, 12818–12823 (2004).
Bostock, M. G., Hyndman, R. D., Rondenay, S. & Peacock, S. M. An inverted continental Moho and serpentinization of the forearc mantle. Nature 417, 536–538 (2002).
Williams, S. E. & Gubbins, D. Origin of Long-Wavelength Magnetic Anomalies at Subduction Zones. J Geophys Res Solid Earth 124, 9457–9473 (2019).
Blakely, R. J., Brocher, T. M. & Wells, R. E. Subduction-zone magnetic anomalies and implications for hydrated forearc mantle. 33, 445–448 (2005).
Hacker, B. R. H2O subduction beyond arcs. Geochem Geophys Geosystems 9, n/a-n/a (2008).
Connolly, J. A. D. Computation of phase equilibria by linear programming: A tool for geodynamic modeling and its application to subduction zone decarbonation. Earth Planet Sc Lett 236, 524–541 (2005).
Holt, A. F. & Condit, C. B. Slab Temperature Evolution Over the Lifetime of a Subduction Zone. Geochem Geophys Geosystems 22, (2021).
Wolery, T. J. EQ3/6, a software package for geochemical modeling of aqueous systems: Package overview and installation guide (Version 7.0). (1992) doi:10.2172/138894.
Huang, F. & Sverjensky, D. A. Extended Deep Earth Water Model for predicting major element mantle metasomatism. Geochim Cosmochim Ac 254, 192–230 (2019).
Boutier, A., Brovarone, A. V., Martinez, I., Sissmann, O. & Mana, S. High-pressure serpentinization and abiotic methane formation in metaperidotite from the Appalachian subduction, northern Vermont. Lithos 396, 106190 (2021).
Bodinier, J.-L. & Godard, M. 2.04 - Orogenic, Ophiolitic, and Abyssal Peridotites. in (eds. Holland", ["Heinrich D. & Turekian"], "Karl K.) 1–73 (Pergamon, Oxford, 2007). doi:10.1016/b0-08-043751-6/02004-1.
Nakatani, T. & Nakamura, M. Experimental constraints on the serpentinization rate of fore-arc peridotites: Implications for the upwelling condition of the slab‐derived fluid. Geochem., Geophys., Geosystems 17, 3393–3419 (2016).
Nakatani, T. & Nakamura, M. Preferential Orthopyroxene Serpentinization and Implications for Seismic Velocity Interpretation in the Fore-Arc Mantle. J Geophys Res Solid Earth 124, 3420–3435 (2019).
Dandar, O., Okamoto, A., Uno, M. & Tsuchiya, N. Mantle hydration initiated by Ca metasomatism in a subduction zone: An example from the Chandman meta-peridotite, western Mongolia. Lithos 452, 107212 (2023).
Lindquist, P. C., Condit, C. B., Hoover, W. F., Hernández-Uribe, D. & Guevara, V. E. Metasomatism and Slow Slip: Talc Production Along the Flat Subduction Plate Interface Beneath Mexico (Guerrero). Geochem., Geophys., Geosystems 24, (2023).
Ely, T. D., Leong, J. M., Canovas, P. A. & Shock, E. L. Huge Variation in H2 Generation During Seawater Alteration of Ultramafic Rocks. Geochem., Geophys., Geosystems 24, (2023).
Hacker, B. R., Abers, G. A. & Peacock, S. M. Subduction factory 1. Theoretical mineralogy, densities, seismic wave speeds, and H2O contents. J Geophys Res Solid Earth 1978 2012 108, (2003).
Evans, B. W. Lizardite versus antigorite serpentinite: Magnetite, hydrogen, and life(?). Geology 38, 879–882 (2010).
Merdith, A. S. et al. Global Hydrogen Production During High-Pressure Serpentinization of Subducting Slabs. Geochem., Geophys., Geosystems 24, (2023).
Mottl, M. J., Komor, S. C., Fryer, P. & Moyer, C. L. Deep-slab fluids fuel extremophilic Archaea on a Mariana forearc serpentinite mud volcano: Ocean Drilling Program Leg 195. Geochem., Geophys., Geosystems 4, (2003).
Barry, P. H. et al. Forearc carbon sink reduces long-term volatile recycling into the mantle. Nature 568, 487–492 (2019).
Young, E. D. et al. The relative abundances of resolved l2CH2D2 and 13CH3D and mechanisms controlling isotopic bond ordering in abiotic and biotic methane gases. Geochim. Cosmochim. Acta 203, 235–264 (2017).
Hyndman, R. D. & Peacock, S. M. Serpentinization of the forearc mantle. Earth Planet. Sci. Lett. 212, 417–432 (2003).
Holland, T. J. B. & Powell, R. An improved and extended internally consistent thermodynamic dataset for phases of petrological interest, involving a new equation of state for solids. J Metamorph Geol 29, 333–383 (2011).
Woodland, A. B., Kornprobst, J. & Tabit, A. Ferric iron in orogenic lherzolite massifs and controls of oxygen fugacity in the upper mantle. Lithos 89, 222–241 (2006).
Ague, J. J. Evidence for major mass transfer and volume strain during regional metamorphism of pelites. Geology 19, 855–858 (1991).
Walters, J. B., Cruz-Uribe, A. M. & Marschall, H. R. Isotopic Compositions of Sulfides in Exhumed High‐Pressure Terranes: Implications for Sulfur Cycling in Subduction Zones. Geochem., Geophys., Geosystems 20, 3347–3374 (2019).
Evans, B. W., Dyar, M. D. & Kuehner, S. M. Implications of ferrous and ferric iron in antigorite. Am. Miner. 97, 184–196 (2012).
Eberhard, L., Frost, D. J., McCammon, C. A., Dolejš, D. & Connolly, J. A. D. Experimental Constraints on the Ferric Fe Content and Oxygen Fugacity in Subducted Serpentinites. J. Pet. 64, egad069 (2023).
Redlich, O. Eine allgemeine Beziehung zwischen den Schwingungs- frequenzen isotoper Molekeln. Zeitschrift für Physikalische Chemie 28B, 371–382 (1935).
Becke, A. D. Density-functional thermochemistry. III. The role of exact exchange. Journal of Chemical Physics 98, 5648–5652 (1993).
Lee, C., Yang, W. & Parr, R. G. Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 37, 785–789 (1988).
Krishnan, R., Binkley, J. S., Seeger, R. & Pople, J. A. Self-consistent molecular orbital methods. XX. A basis set for correlated wave functions. J. Chem. Phys. 72, 650–654 (1980).
Blaudeau, J.-P., McGrath, M. P., Curtiss, L. A. & Radom, L. Extension of Gaussian-2 (G2) theory to molecules containing third-row atoms K and Ca. Journal of Chemical Physics 107, 5016–5021 (1995).
Hill, P. S., Tripati, A. K. & Schauble, E. A. Theoretical constraints on the effects of pH, salinity, and temperature on clumped isotope signatures of dissolved inorganic carbon species and precipitating carbonate minerals. Geochim. Cosmochim. Acta 125, 610–652 (2014).
Frisch, M. J. et al. Gaussian 16 Rev. C.01. (2016).

There is NO Competing Interest.

SupplementaryInformationfinal.pdf

Download PDF

Version 1

posted

You are reading this latest preprint version

Decoupling of high-pressure H₂ production from serpentinization and magnetite in subduction zones

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Methods

Declarations

Data availability

Competing Interests

Author Contributions

Acknowledgments

References

Additional Declarations

Supplementary Files

Status:

Version 1