Massive abiotic methane production in eclogite during cold subduction

Abstract

Methane (CH₄) is a critical but overlooked component in the study of the deep carbon cycle. Abiotic CH₄ produced by serpentinization of ultramafic rocks has received extensive attention, but its formation and flux in mafic rocks during subduction remain poorly understood. Here, we report massive CH₄-rich fluid inclusions in well-zoned garnet from eclogites in Western Tianshan, China. Petrological characteristics and carbon–hydrogen isotopic compositions confirm the abiotic origin of this CH₄. Reconstructed P–T–fO₂–fluid trajectories and Deep Earth Water modeling imply that massive abiotic CH₄ was generated during cold subduction at depths of 50–120 km, whereas CO₂ was produced during exhumation. The massive production of abiotic CH₄ in eclogites may result from multiple mechanisms during prograde high pressure-ultrahigh pressure metamorphism. Our flux calculation proposes that abiotic CH₄ that has been formed in HP-UHP eclogites in cold subduction zones may represent one of the largest, yet overlooked, sources of abiotic CH₄ on Earth.

Massive abiotic methane can be produced during prograde HP-UHP metamorphism in cold subduction zone, which may represent one of the largest sources of abiotic CH₄ on Earth.

INTRODUCTION

Methane (CH₄) can form on Earth via biotic or abiotic processes. The existence of abiotic CH₄ has important implications for deep subsurface microbial life, natural gas exploration, global warming and the long-term habitability of Earth [1,2]. Abiotic CH₄ widely exists in Precambrian crystalline shields, volcanic and geothermal systems, inclusions in crystalline intrusions and serpentinized ultramafic rocks in submarine peridotite-hosted hydrothermal systems, continental ophiolites and peridotite massifs [1–6]. Both experiments [7–9] and theoretical calculations [10,11] indicate that abiotic CH₄ can form and exist under pressure–temperature (P–T) conditions corresponding to the Earth's upper mantle. Recent experiments have shown that abiotic CH₄ can be produced in the presence of aqueous fluids at 1.5–3.5 GPa and 300–700°C, P–T conditions that may occur in cold subduction zones [12–14]. Indeed, abiotic CH₄ has been discovered in fluid inclusions in metasomatized ultramafic ophicarbonates from the cold subduction zones in the Alps [15,16] and Western Tianshan [17], where it formed during serpentinization at forearc depths. However, the formation and pathways of abiotic CH₄ produced in mafic rocks, especially in eclogites, in natural subduction zones are still unknown.

Carbon is transported from Earth's surface to the mantle in subduction zones [18]. Some of this recycled carbon may be released as CH₄ and CO₂ via metamorphism or volcanism. It is controversial whether the vast majority of oceanic carbon is released at forearc to subarc depths [19–22] or deeply subducted [23]. In addition, the amounts of CH₄ stored in the subducting slab, introduced into the overriding mantle wedge and emitted to the atmosphere through arc magmatism, are poorly constrained. Thus, lacking understanding of the geological, physical and chemical conditions of deep abiotic CH₄ formation prevents the quantification of these problems.

The Western Tianshan ultrahigh pressure (UHP) metamorphic belt in China, the world's largest cold oceanic subduction zone [24], provides one such case for unraveling the fate of abiotic CH₄ formed in eclogite at forearc to subarc depths during cold subduction. Coesite relics in various lithologies confirm the estimates of peak metamorphic conditions of 2.7–3.2 GPa and ∼550°C, corresponding to subduction depths of 90–110 km [24]. These rocks represent the most deeply subducted remnant of the Paleo-Asian oceanic lithosphere during the Carboniferous, which is currently exposed along the southern part of the Central Asian Orogenic Belt [24]. Previous petrological and experimental studies showed that CH₄-bearing fluid inclusions in omphacite from Western Tianshan high pressure (HP) eclogites were formed by the reduction of Fe²⁺-bearing carbonates under low oxygen fugacity [25]. Herein, we first report massive CH₄-rich primary fluid inclusions in well-zoned prograde garnet from UHP eclogites from the Western Tianshan metamorphic belt, which allow us to determine the P–T–fO₂ conditions of abiotic CH₄ formation and to estimate the CH₄ flux released from eclogites during subduction.

RESULTS

Prograde metamorphic garnets

A CH₄-rich carbonated eclogite sample (HB142-8) that retains prograde metamorphic mineral assemblages was chosen from >30 samples collected from Western Tianshan (Supplementary Fig. S1). The garnets are Fe-rich, have an almandine molar fraction (X_alm = Fe²⁺/[Ca + Mg + Fe²⁺ + Mn]) of 52%–73%, and display complex chemical zoning (Fig. 1a; Supplementary Fig. S2 and Table S1). Most garnets show four growth zones: a core region (Grt I) with a high grossular molar fraction (X_grs = Ca/[Ca + Mg + Fe² + Mn]) of 15%–22%; an inner mantle (Grt II) that has the lowest X_grs of 11%–15%; an outer mantle (Grt III) where X_grs increases to 22%–26%; and a rim (Grt IV), which has the highest X_grs of 29%–31%. The chemical zoning of X_alm is opposite to that of X_grs. The pyrope molar fraction (X_py = Mg/[Ca + Mg + Fe²⁺ + Mn]) only increases slightly from Grt I to Grt IV, while the spessartine molar fraction (X_sps = Mn/[Ca + Mg + Fe²⁺ + Mn]) decreases. This chemical variation in garnet reveals prograde metamorphism during subduction. Single and multiphase solid inclusions such as graphite, Mg-calcite, omphacite, chlorite, lawsonite and rutile commonly occur in Grt I and Grt II. In Grt III and Grt IV, only a few rutile, omphacite, lawsonite and its pseudomorph are present (Fig. 1a; Supplementary Figs S2 and S3). Omphacite also shows zonation from core to rim, with FeO contents decreasing gradually from 7.77 to 2.85 wt% (Fig. 1a; Supplementary Table S1).

Figure 1.

Examples of CH₄-rich fluid inclusions in eclogite from the Western Tianshan subduction zone. (a) An Fe X-ray map showing the four well-preserved growth zones of garnet (Grt I–IV), abundant CH₄-rich fluid inclusions preserved in Grt I and Grt II, and the paucity of fluid inclusions in Grt III and Grt IV. (b) Abundant tubular two- or three-phase CH₄-rich fluid inclusions parallel with the long axis of omphacite (Omp). (c–j) Examples of two- or three-phase CH₄-rich fluid inclusions in (c–e) Grt I, (f–h) Grt II and (i–j) Omp; gas phases are CH₄ with minor N₂, liquid phase is H₂O, and solid phases (if present) are calcite (Cal) and graphite (Gr). Note that the representative locations of the CH₄-rich fluid inclusions are marked in numbered squares in (a). Supplementary Figs S2–S6 provide additional element maps, microphotos and locations of CH₄-rich fluid inclusions.

CH₄-rich fluid inclusions

Abundant CH₄-rich fluid inclusions are well preserved in Grt I, Grt II and in cores and mantles of omphacite (Omp), but are not present in Grt III, Grt IV and Omp rims (Fig. 1; Supplementary Figs S4–S6). Fluid inclusions in Grt I are mostly elliptic, equiaxial or short columnar (Fig. 1c–e), whereas those in Grt II are mostly tubular, prismatic, dendritic or irregular along with the zoning, and commonly larger than those in Grt I (Fig. 1f–h; Supplementary Fig. S4a–d). Fluid inclusions in Omp occur as isolated parallel tubes or as intragranular clusters, with their long dimensions parallel to the c-axes of the host Omp (Fig. 1b, i and j; Supplementary Fig. S4e and f). These features unequivocally demonstrate the primary nature of the fluid inclusions. These fluid inclusions are different from the secondary fluid inclusions, which were enclosed by olivine during dissolution–precipitation and whose CH₄ formed due to rarely internal post-entrapment serpentinization [6,26]. Through extensive investigations of samples taken across the Western Tianshan HP-UHP metamorphic belt, we observed that CH₄-rich fluid inclusions are commonly located in the core and mantle of garnet and omphacite in eclogite, although their prograde zonings are not preserved as well as in the sample HB142-8.

The CH₄-rich fluid inclusions in garnet (Grt I and Grt II) and omphacite contain liquid + vapor ± solid daughter crystals at room temperature. The solids are generally very tiny with irregular shapes. The ubiquitous CH₄ vapor has sharp Raman peaks ranging from 2916 to 2918 cm⁻¹, centered mainly at 2917 cm⁻¹, and ubiquitous N₂ marked by weak bands from 2331 to 2328 cm⁻¹ (Supplementary Note 1). The fluid phase has the characteristic broad Raman peak for H₂O at 3440 cm⁻¹ (Fig. 2a–c; Supplementary Figs S5 and S6). Daughter crystals are mostly calcite and graphite, and less commonly rutile. The larger solid with a polygonal–prismatic shape has high birefringence and is identified as Mg-calcite with Raman peaks at 1087, 712 and 281 cm⁻¹ (Fig. 2d). Most Mg-calcites are very tiny, and have only weak peaks at 1087 cm⁻¹. The strong Raman peaks centered at 1583 and 1356 cm⁻¹ indicate the presence of well-crystallized graphite (Fig. 2a; Supplementary Figs S5 and S6). Abundant CO₂-rich fluid inclusions, which are mostly larger than the CH₄-rich fluid inclusions, are present in retrograde ankeritic dolomite (for details see Supplementary Note 1 and Fig. S7). The CO₂ vapor has two Raman peaks at 1285 and 1386 cm⁻¹; the fluid phase also has the broad Raman peak for H₂O at 3440 cm⁻¹ (Fig. 2e; Supplementary Fig. S8). No H₂ was detected in any of the fluid inclusions.

Figure 2.

Examples of Raman spectra and hyperspectral confocal Raman maps of CH₄-rich fluid inclusions. (a–c) CH₄-rich fluid inclusions in Grt I, Grt II and Omp, respectively. (d) Calcite daughter crystal in a fluid inclusion in Omp. (e) CO₂-rich fluid inclusion in retrograde ankeritic dolomite. ‘Grt/Omp/Dol BG’ mark the background peaks of the host garnet, omphacite or dolomite, respectively. The inset images are photomicrographs and their corresponding hyperspectral confocal Raman maps of the fluid inclusions, showing the existence of CH₄ and CO_2. White dashed lines outline the inclusions.

Stable C-H isotopic compositions of CH₄

The δ¹³C and δ²H values of CH₄ (see Methods) in both garnet and omphacite from the Western Tianshan eclogites are rather homogeneous, relatively ¹³C enriched and ²H depleted. The δ¹³C values of CH₄ in garnet vary from −30.9‰ to −28.6‰ and δ²H values span −383.0‰ to −363.1‰. The δ¹³C and δ²H values of CH₄ in omphacite are −30.7‰ to −29.3‰ and −375.6‰ to −359.5‰, respectively (Fig. 3; Supplementary Table S2). Our δ¹³C and δ²H data plot entirely in the field of potentially abiotic CH₄ origin, none overlapping with the microbial, volcanic and sedimentary thermogenic CH₄ fields (Fig. 3) [27].

Figure 3.

Stable C and H isotope compositions of CH₄. (a) The global summary of CH₄ with outlines of the microbial, sedimentary thermogenic, volcanic thermogenic and potentially abiotic areas taken from reference [27]. New data for CH₄ gas preserved in eclogites from Western Tianshan plot in the field of potentially abiotic origin. Abiotic CH₄ data for serpentinization, inclusions in crystalline intrusions and Precambrian shields are shown for comparison [1]. (b) The enlargement of the rectangle in (a) clearly shows the δ¹³C and δ²H values in both garnet and omphacite from the eclogite samples from the Western Tianshan.

Phase equilibrium modeling and oxygen fugacity calculation

The growth zoning of garnet from the Western Tianshan eclogite is revealed to record an amazing P–T–fO₂ trajectory (Fig. 4), which we quantify with phase equilibrium modeling and fO₂ calculation (see Methods, Supplementary Fig. S9 and Tables S3–S5). Grt I reflects prograde metamorphism at 2.1–2.3 GPa, 500–530°C (∼70 km), with 2.4 log units below the Fayalite–Magnetite–Quartz buffer (FMQ). Grt II shows UHP peak metamorphism at 3.2–3.4 GPa, 540–560°C (∼110 km), in accord with the ubiquity of coesite in the same unit [24], and the fO₂ decreases to the lowest FMQ value of −3.5. The temperature estimates are consistent with the results of 530–560°C obtained through Raman spectra of graphite coexisting with CH₄ [28]. Grt III and Grt IV reveal heating with decompression to 2.7–2.8 GPa at 560–580°C, and 2.3–2.5 GPa at 570–590°C, respectively; the fO₂ increases to FMQ of −2.2 and −1.0 for Grt III and Grt IV, respectively.

Figure 4.

P–T–fO₂-fluid evolution trajectory of the representative CH₄-rich eclogite HB142-8. Circles with I to IV are P–T conditions corresponding to the four growth zones of garnet obtained by Perple_X in the system of MnNCKFMASCHO (Supplementary Fig. S9). The blue and orange lines with an arrow represent the prograde and retrograde HP-UHP metamorphic path, respectively. The marked oxygen fugacity (fO₂) values were calculated by paired garnet and corresponding omphacite inclusions captured in Grt I–IV (Supplementary Tables S1, S3 and S4).

Metamorphic aqueous fluid composition by DEW modeling

In order to quantify the compositional evolution of aqueous fluid in the Western Tianshan subduction zone, the DEW (Deep Earth Water) model [29,30] was used (see Methods). The metamorphic fluid compositions largely depend on the P–T–fO₂ conditions (Fig. 5). Carbon species are dominated by reduced aqueous CH₄, which increased from 61% at ∼50 km (Fig. 5a) to 97% at ∼80 km (Fig. 5b) and ∼120 km (Fig. 5c), along with prograde HP-UHP metamorphism. This is consistent with our observation of massive CH₄ in prograde Grt I–II and Omp (Fig. 1 and Supplementary Figs S4–S6). The abiotic CH₄ actually began to form in the blueschist facies at depths of <50 km, which is not recorded by garnet. By contrast, the amount of CH₄ suddenly decreased during the decompression and exhumation, while the proportion of oxidized carbon species, such as CO₂ and H₂CO₃, gradually increased (Fig. 5d–f). This agrees with the absence of CH₄ in Grt III and Grt IV, and the presence of CO₂ in retrograde ankeritic dolomite (Supplementary Figs S7 and S8), although we did not detect other oxidized carbon species, such as H₂CO₃ and CH₃CH₂COO⁻. Consequently, the calculated fluid composition is basically consistent with our observation.

Figure 5.

Compositional variation of aqueous carbon species in eclogitic fluid along the P–T–fO₂-fluid evolution trajectory in Fig. 4. (a–c) Carbon species are dominated by reduced aqueous CH₄ during the cold subduction (prograde HP-UHP metamorphism). (d–f) The proportion of oxidized carbon species gradually increases during the decompression and exhumation.

DISCUSSION AND IMPLICATIONS

The variation of carbon species and the redox properties in subduction-zone metamorphic fluids

The combination of the petrological observations (Figs 1 and 2 and Supplementary Figs S1–S8) and thermodynamic models (Figs 4 and 5 and Supplementary Fig. S9) leads to a complete P–T–fO₂-fluid evolution trajectory for the CH₄-bearing eclogites. During the prograde metamorphism from the blueschist facies at ∼50 km to eclogite facies at ∼70 km (Grt I), the carbon species in the fluids were dominated by reduced CH₄, and the proportion of CH₄ increased from ∼61% to ∼97%. From HP eclogite facies at ∼70 km (Grt I) to the UHP eclogite facies at ∼120 km (Grt II), the carbon species in the aqueous fluid was mainly CH₄, and it remained at the maximum of ∼97%. This is consistent with the behavior of sulfur in the same area [31]: reduced H₂S and HS⁻ prevailed in the fluids, and their release reached its climax at ∼90 km depth [31]. During the exhumation with heating and decompression (Grt III–Grt IV), the fluid evolved into a CO₂–CH₄-bearing aqueous fluid and the oxygen fugacity increased to FMQ-1. This is consistent with the rare presence of fluid inclusions with coexisting CO₂ and CH₄ in the core of a retrograde ankeritic dolomite (Supplementary Note 1 and Fig. S8h). With the progression of the decompression and the increase of the oxygen fugacity, the carbon species in the fluids completely changed into CO₂, and CO₂ perpetuated through the epidote–amphibole facies to even shallower depths during the exhumation, as recorded by CO₂-rich fluid inclusions in the rim of the retrograde ankeritic dolomite (Fig. 2e; Supplementary Figs S7 and S8). Therefore, our study provides new clues with regard to the controversy over the reduced [31,32] or oxidized [33–36] state of the subduction zone fluids. Our study demonstrates that the compositions and the redox properties of subduction zone fluids vary with the P–T–fO₂ conditions, and the fluids are neither always oxidized nor perpetually reduced. Therefore, it is necessary to characterize the fluid compositions and properties for each distinct metamorphic stage in subduction zones.

The mechanisms and origin of CH₄

The reduction of Fe²⁺-bearing carbonates under low oxygen fugacity is an important mechanism for the formation of abiotic CH₄ in subduction zones, and experiments reproduced it [25]. Textures imply that such a metamorphic reaction (more precisely, the decomposition of relict prograde ankeritic dolomite to calcite/aragonite, magnetite and graphite) proceeded in the samples studied herein (Supplementary Note 1 and Supplementary Fig. S7a). This substantiates that the reaction ankeritic dolomite + H₂O => calcite/aragonite + magnetite + graphite + CH₄ was the main reaction for CH₄ production in the Western Tianshan eclogites. In addition, the inclusions of graphite, Mg-calcite, lawsonite and chlorite in the prograde Grt I and Grt II (Supplementary Fig. S2 and S3) point to another possible reaction for CH₄ production in the Western Tianshan eclogites: graphite + lawsonite +/− chlorite = > calcite/Mg-calcite + epidote + CH₄ + H₂O. This reaction was verified by experiments, in which the abiotic hydrocarbons were produced by the reduction of graphite or diamond with aqueous fluids at high-P and low-T conditions [12,37–39]. Moreover, carbonate dissolution is considered as the main decarbonization mechanism in subduction zones [19,34,35]. We cannot exclude the fluid-mediated dissolution and precipitation of carbonates because of the common occurrence of calcite in the fluid inclusions (Figs 1b, j, 2b–d and Supplementary Figs S5 and S6). Therefore, besides the reduction of Fe²⁺-bearing carbonates, metamorphic reactions of graphite and carbonate dissolution may also be efficient mechanisms that produced abiotic CH₄ during the prograde HP-UHP metamorphism in the Western Tianshan eclogites. By contrast, during the exhumation, the increase of oxygen fugacity favors the production of CO₂: CH₄ released by Grt I, Grt II and Omp was converted into CO₂ subsequently, which was captured by retrograde ankeritic dolomite while fO₂ increased (Supplementary Figs S7 and S8).

The petrographic textures and the mechanisms for the formation of CH₄ outlined above do not involve organic carbon, pointing to an abiotic origin of the CH₄ in eclogites. Moreover, our δ¹³C and δ²H data plot into the field of abiotic origin, but they are different from the abiotic CH₄ formed in crystalline intrusions, Precambrian shields and serpentinization (Fig. 3) [1,27]. Reversely, if the CH₄ in our inclusions has a biotic origin, the hydrogen isotope composition of CH₄ and H₂O should have re-equilibrated during entrapment. Grt II recorded a temperature of up to 560°C (Fig. 4), and δ²H of the coexisting H₂O is estimated at about −270‰ [40]. Such a depleted hydrogen isotope composition in a subduction zone fluid system is not expected [41]. The same scenario applies to carbon isotopes. Hydrogen and carbon isotope compositions are in disequilibrium with the surrounding hydrogen and carbon sources, indicating the CH₄ formed by abiotic origin. Therefore, the petrological characteristics and C–H isotopic compositions both confirm the abiotic origin of the CH₄ in this study.

Cold subduction boosts abiotic CH₄ production

Abiotic CH₄ has increasingly been recognized in HP-UHP metamorphic rocks from subduction zones. CH₄-rich fluid inclusions occur in high pressure-low temperature (HP-LT) ophicarbonates from the Western Alps [15,16], the Alpine Corsica [16] and the Western Tianshan [17]; in HP/UHP-LT eclogites from the Western Tianshan [25 and this study]; in metaperidotites from the Appalachian [42] and the North Qilian HP-LT metamorphic belt [43]; and in continental-type HP/UHP-MT eclogites (but solely in local reduced environments) [44,45]. From these examples it seems that cold subduction zones with HP/UHP-LT conditions favor the production of abiotic CH₄.

The thermal structure differs distinctly between cold and warm subduction zones [46]. For example, cold subduction zones are characterized by much deeper serpentinization (∼100 km depth) than warm subduction zones (<35 km) [16]. By contrast, the serpentinization of the mantle wedge is predicted to be more pronounced in warm subduction zones compared to cold ones [16,47]. These conditions favor high H₂-CH₄ concentrations in fluid in cold subducting slabs, but high H₂ and CH₄ fluxes in the mantle wedge in warm subduction zones [16]. Consequently, cold subducting slabs are more reduced and advantageous for CH₄ production than warm subducting slabs at depths of 35–100 km; this concurs with the massive CH₄ production in eclogites from forearc to subarc depth observed herein (Fig. 4). Concordantly, most recent experiments indicate that CH₄-bearing reduced fluid can be produced at P–T conditions (1.5–3.5 GPa and 300–700°C) comparable to cold subduction zones [12–14]. Experiments and simulation calculations also predicted that the production of abiotic CH₄ is facilitated at low-temperature conditions (T < 1500 K), whereas dissociation to higher hydrocarbons proceeds at high-temperature conditions (T > 1500 K) [7–11,39]. The DEW model calculation also shows that CH₄ has a much higher proportion than CO₂ in aqueous eclogitic fluids in cold subduction zones (T < 700°C) compared to warm subduction zones (T > 700°C) at fixed conditions of 5 GPa and FMQ-2 [48]. Thermodynamic calculations also demonstrated that CH₄ is stabilized in cold subduction zones relative to CO₂, especially in graphite-saturated C-O-H systems like the one studied herein [37]. All this evidence indicates that cold subduction zones are more conducive to the formation of abiotic CH₄ than warm subduction zones. In fact, the fO₂ of the subducted slabs are remarkably heterogeneous due to varying degrees of alteration at the seafloor, and fluid/rock ratios and redox budgets should be considered [36,37,49,50]. Consequently, we conclude that cold subduction zones can boost abiotic CH₄ production at favorable fO₂ conditions and bulk rock compositions.

Abiotic methane flux released from HP-UHP eclogites during cold subduction

Our data permit provisional estimates of the abiotic CH₄ flux released from eclogites in worldwide modern subduction zones. Based on the moles of CH₄ per kilogram of H₂O calculated with the DEW model at specific P–T–fO₂ conditions during prograde metamorphism from 50 km to 120 km (Fig. 5a–c; Supplementary Table S6), the amount of H₂O released from the subducting slab along our prograde P–T–fO₂ trajectory (∼4 wt%) [51] and the total mass of eclogites subducted annually, the calculated CH₄ flux released from eclogites could be 10.85 Mt/y (Supplementary Note 2 and Table S7). We also considered the amount of H₂O lost annually from global mafic rocks from depths of 50 km to 100 km (2.865 × 10¹⁴ g/y) [46]: the multiplication of this value with the average moles of CH₄ per kilogram of H₂O (2.353 mole/kg) yields a total released CH₄ flux of 10.79 Mt/y. The estimated results obtained from the two different methods are nearly the same, which indicates that the results are reliable. The CH₄ flux released from the mafic eclogites (∼10.8 Mt/y) much exceeds the abiotic CH₄ production at mid-ocean ridges (1.1–1.9 Mt/y) [52] and by HP serpentinization worldwide (2.3 × 10⁻³ to 1 Mt/y) [16].

We have discussed that the reduced and cold subduction zones favor the production of abiotic CH₄ (see above). Therefore, to conservatively estimate abiotic CH₄ flux released from eclogites during cold subduction, we also considered cold paleo-subduction zones with reduced environments. In the case of the ancient Southern Tianshan cold subduction zone, the CH₄ flux released from the eclogites during the prograde HP-UHP metamorphism at depths of 50–120 km is 0.49 Mt/y (Supplementary Note 2 and Table S7). The subduction-related production of CH₄ ± H₂ has been recorded in ophicarbonates and metaperidotites from the Alps, Western Tianshan, North Qilian and Appalachian at FMQ −6.0 to FMQ −2.0 [15–17,42,43], which is within the range of FMQ-2 to FMQ-4.5 known to permit the stability of CH₄ [37,48,50]. Eclogites in such reduced and cold subduction zones [15–17,42,43] are capable of producing abiotic CH₄, and the estimated CH₄ flux could be 1.76 Mt/y (Supplementary Note 2 and Tabl  S7). The above estimates do not include either abiotic CH₄ from the HP serpentinization [15–17], or the potential contribution from the metapelitic schists that experienced similar P–T–fO₂ paths as the HP-UHP eclogites [24]. Consequently, the subducted cold oceanic crust may be an important, yet overlooked, source of abiotic CH₄, which cannot be ignored in estimates of the global carbon flux. The released abiotic CH₄ might migrate upwards and affect the redox state of the overlying mantle wedge. Potentially, it contributes to natural gas deposits at shallow depths, and/or returns to the atmosphere by degassing through arc volcanoes, further influencing the climate and the environment (Fig. 6).

Figure 6.

Illustration of the abiotic CH₄ production during prograde HP-UHP metamorphism in an oceanic subduction zone (a). The subduction zone comprises oceanic crust, slab mantle, mantle layers and continental crust. On the right, the images (b and c) show the entrapment of CH₄-rich fluid inclusions during the growth of prograde Grt I, Grt II and Omp, corresponding to P–T–fO₂ conditions during deep subduction (enlarged image in the middle). The abiotic CH₄ produced during the prograde HP-UHP metamorphism migrates upwards to shallow depths, potentially contributing to shallow gas or oil deposits or returning from subarc depths to the surface by degassing from arc volcanoes.

METHODS

Major element compositions and Raman spectroscopy analysis

Major element and Raman spectroscopy analysis were both conducted at Peking University. The former was performed on a JEOL JXA-8230 electron microprobe using a 15 kV acceleration voltage and 10 nA beam current. The beam diameter was set to 2 μm for silicates and 10 μm for carbonates. Raman spectroscopy was performed on a HORIBA Jobin Yvon confocal LabRAM HR Evolution micro-Raman system equipped with a frequency doubled green Nd-YAG laser (532 nm), a 100× short-working distance objective, and a stigmatic 800 mm spectrometer with a 600 groove/mm grating. The laser spot size was focused to 1 μm and the accumulation time varied between 60 and 120 s. The estimated spectral resolution was greater than 1.0 cm⁻¹ and the calibration used synthetic silicon. Hyperspectral Raman images were collected along a regular grid of points, nearly equidistant in both directions, with a computer-controlled, automated X–Y mapping stage. All Raman data are presented in the original raw spectra state, without any baseline subtraction correction.

Carbon and hydrogen isotope analyses

The δ¹³C and δ²H values are expressed as δ¹³C (‰) = [(¹³C/¹²C)_sample/(¹³C/¹²C)_VPDB − 1] × 1000, and δ²H = [(²H/¹H)_sample/(²H/¹H)_VSMOW − 1] × 1000, respectively. VSMOW is Vienna Standard Mean Ocean Water and VPDB is Vienna Pee Dee Belemnite. The δ¹³C and δ²H values of abiotic CH₄ from Western Tianshan eclogites were determined using a continuous flow isotope ratio mass spectrometry technique with a Thermo Scientific gas chromatograph (GC) and Thermo Finnigan MAT253 isotope ratio mass spectrometer at the Analytical Laboratory, Beijing Research Institute of Uranium Geology, China. Samples of the host minerals (pure garnet or omphacite) were first washed with diluted HCl in order to remove carbonate, and then washed repeatedly with deionized water (DI water) to remove any contaminants from the crystal surfaces. Methane was extracted from the samples by thermally decrepitating the fluid inclusions. This involved heating the samples for 15 min with helium gas at 550°C, passing the released gas through a NaOH trap and a Mg(ClO₄)₂ trap to remove CO₂ and H₂O, and then through a liquid N₂ cold trap to enrich CH₄. Trace amounts of interfering compounds were separated by gas chromatography after pre-concentration of the CH₄ sample. The purified sample was then either combusted to CO₂ by reaction with CuO, or pyrolyzed to H₂ in a silica tube that was heated to 1420°C prior to mass spectrometry. The chromatographic column was PoraPLOT Q (27.5 m × 0.32 mm × 0.10 μm). The δ¹³C and δ²H values were corrected by three methane C–H isotope standards, IsoRM 201, 202 and 203 (Qingdao IsotopTech Ltd). The results of the standard samples are consistent with the standard values within error (Supplementary Table S8). The precisions (RSD) of δ¹³C and δ²H were better than 0.3‰ and 3‰, respectively. In addition, the continuous flow isotope ratio mass spectrometry technique can quickly carry the released gases out of the heat source and minimize the possible isotopic fractionation. The peaks of the GC spectrum of CH₄ in both garnet and omphacite are perfect (Supplementary Fig. S10), without any tails, which highlights the good quality of our data.

Thermodynamic phase equilibrium modeling and oxygen fugacity calculation

The P–T phase diagram (Fig. 4) for the eclogite sample HB142-8 was calculated in the MnNCKFMASCHO system using Perple_X software [53,54] with an internally consistent thermodynamic data set [55]. The details are shown in Supplementary Fig. S9. The Fe³⁺/ΣFe ratio in garnet b, which contained omphacite inclusions in Grt III (Supplementary Fig. S2b), was measured by the flank method [56] with the JEOL JXA-8100 electron microprobe at the Key Laboratory of Orogenic Belts and Crustal Evolution, School of Earth and Space Sciences, Peking University, China. The analytical procedure of the flank method followed reference [57] in detail. The results corresponding to the four zones of garnet (Grt I– IV) are shown in Supplementary Table S3. The oxygen fugacity (fO₂) calculation was calculated for Grt I–IV, and the corresponding omphacite with the equilibrium reaction: 1/3 Ca₃Al₂Si₃O₁₂ (garnet) + 5CaFeSi₂O₆ (clinopyroxene) + O₂ = 2Ca₃Fe₂Si₃O₁₂ (garnet) + 1/3Fe₃Al₂Si₃O₁₂ (garnet) + 4SiO₂ (coesite) [58]. To relate the results with the P–T conditions obtained with phase equilibrium modeling (Supplementary Fig. S9), we determined fO₂ at 520°C and 22.5 kbar, 550°C and 33 kbar, 570°C and 27.5 kbar and 580°C and 23 kbar, corresponding to Grt I–IV, respectively. Due to the paucity of analyzable omphacite inclusions corresponding to Grt I–II and Grt IV in garnet b, we paired omphacite inclusions from another two garnets, c and d, with Grt I and Grt II, and matrix Omp rim with Grt IV (Supplementary Fig. S3 and Table S1). The compositions of the omphacite inclusions in Grt I–III, and matrix omphacite, are listed in Table S1. The calculated fO₂ are listed in Supplementary Table S4.

DEW calculation of carbon concentration in fluids

The DEW model [29,30] enables the calculation of reaction equilibrium constants involving minerals, aqueous inorganic and organic ions, complexes, and neutral species. These equilibrium constants combined with the EQ3 fluid speciation code can be used to calculate the aqueous speciation of a fluid in equilibrium with a given mineral assemblage at fixed fO₂, P and T. We selected six points along the P–T–fO₂ trajectory obtained within Fig. 4 to calculate the C-species compositions (Fig. 5 and Supplementary Table S6). Figure 5b–e corresponds to the P–T–fO₂ conditions recorded by Grt I to Grt IV (Fig. 4), except for the pressure: the DEW model was spaced at 5 kbar intervals, and we chose the nearest integer pressure values. The P–T conditions of Fig. 5a and Fig. 5f followed the prograde and retrograde path and the corresponding mineral assemblage deduced in Fig. 4, respectively; fO₂ accorded with reference [25] (450°C, 1.5 GPa, FMQ-1; 550°C, 2.0 GPa, FMQ + 1). The composition of the main solid solutions was set based on the mineral compositions analyzed in the sample. Thus, the molality of carbon in the fluid was not artificially set, but was to be self-consistent based on the dissolved carbon-bearing minerals for the respective metamorphic stages (Fig. 5a–f). The results and input data files are listed in Supplementary Tables S6 and S9. The method used to estimate the abiotic CH₄ flux is described in Supplementary Note 2.

FUNDING

This work was supported by the National Key Research and Development Program of China (2019YFA0708501) and the National Natural Science Foundation of China (42172060 and 41702052).

Conflict of interest statement. None declared.