Rapid microbial methanogenesis during CO₂ storage in hydrocarbon reservoirs

Abstract

Carbon capture and storage (CCS) is a key technology to mitigate the environmental impact of carbon dioxide (CO₂) emissions. An understanding of the potential trapping and storage mechanisms is required to provide confidence in safe and secure CO₂ geological sequestration^1,2. Depleted hydrocarbon reservoirs have substantial CO₂ storage potential¹,³, and numerous hydrocarbon reservoirs have undergone CO₂ injection as a means of enhanced oil recovery (CO₂-EOR), providing an opportunity to evaluate the (bio)geochemical behaviour of injected carbon. Here we present noble gas, stable isotope, clumped isotope and gene-sequencing analyses from a CO₂-EOR project in the Olla Field (Louisiana, USA). We show that microbial methanogenesis converted as much as 13–19% of the injected CO₂ to methane (CH₄) and up to an additional 74% of CO₂ was dissolved in the groundwater. We calculate an in situ microbial methanogenesis rate from within a natural system of 73–109 millimoles of CH₄ per cubic metre (standard temperature and pressure) per year for the Olla Field. Similar geochemical trends in both injected and natural CO₂ fields suggest that microbial methanogenesis may be an important subsurface sink of CO₂ globally. For CO₂ sequestration sites within the environmental window for microbial methanogenesis, conversion to CH₄ should be considered in site selection.

Main

A possible method for reducing current greenhouse gas emission rates is carbon capture and storage (CCS). Long-term carbon dioxide (CO₂) trapping mechanisms include structural or stratigraphic trapping in stable geological configurations, dissolution into pore fluids (solubility trapping)⁴, carbonate mineralization (precipitation and mineral trapping)⁵ or adsorption (for example, onto coal)⁶. Typically, during CO₂ enhanced oil recovery (CO₂-EOR), a proportion of the injected CO₂ remains within the reservoir post-injection³, providing an analogue for investigating and quantifying processes within CCS sites, over decadal timescales.

Here we investigate the behaviour of CO₂ within the Olla Oil Field, Louisiana, USA (Fig. 1), which was CO₂-EOR flooded in the 1980s, and compare its (bio)geochemical composition with the adjacent Nebo-Hemphill Oil Field, which was never subjected to CO₂-EOR (Methods). Approximately 9 × 10⁷ m³ (standard temperature and pressure (STP)) of injected CO₂ has been retained in the Olla reservoirs post-injection⁷. The injected CO₂ was sourced from the Black Lake Oil Field, which is located adjacent to the Sabine Island Complex, a basement high and likely conduit for mantle-derived fluids⁸. Although the Olla Oil Field has been geochemically characterized by several previous studies^9,10,11 (Methods), this study integrates noble gas and clumped isotope data with stable isotope and microbiological data to investigate the fate of the injected CO₂. Previous studies have suggested that microbial hydrocarbon degradation and methanogenesis may occur within both the Olla and Nebo-Hemphill oil fields^9,10,11. Microbial methanogenesis takes acetate, methylated compounds, or hydrogen and CO₂ (hydrogenotrophic methanogenesis, equation (1)) as its starting point; the latter mechanism is most relevant to this study. Hydrogenotrophic methanogenesis results in ¹³C enrichment of residual CO₂ and the formation of methane (CH₄) initially ¹³C depleted compared with the source CO₂, which increases to the value of the consumed CO₂.

$${{\rm{CO}}}_{2}+4{{\rm{H}}}_{2}\to {{\rm{CH}}}_{4}+2{{\rm{H}}}_{2}{\rm{O}}$$

(1)

Fig. 1: Map of study area showing locations of the Olla and Nebo-Hemphill oil fields as well as the Black Lake Oil Field, from which the injected CO₂ was sourced.

Only the Olla Oil Field contains injected CO₂. The inset on the right shows an expanded view with individual sample locations (nearby urban areas are denoted by stars) and a stratigraphic column showing the relevant lithologic units⁹.

Given their proximity and comparable geological histories (including hydrocarbon production from similar reservoirs; Methods), we assume that the Nebo-Hemphill and Olla oil fields initially had comparable geochemical compositions, before CO₂-EOR at Olla. For example, the Olla and Nebo-Hemphill oil fields have broadly overlapping pH, temperatures, salinities and ion chemistry (Extended Data Table 1). Despite CO₂ injection ceasing in 1986, CO₂ concentrations within the Olla Oil Field (18 ± 12 mol% (STP) across all reservoirs) are greater than breakthrough concentrations limits (10%)⁷ and significantly greater than those measured at Nebo-Hemphill (1.02 ± 0.69 mol% (STP)). We observe a higher δ¹³C_VPDB of CO₂ at Olla (13.5 ± 3.4‰) compared with both Nebo-Hemphill (4.8 ± 4.9‰) and the injected CO₂ (0.85 ± 0.92‰)¹⁰ (Extended Data Table 2), where δ¹³C = [(¹³C/¹²C)_sample/(¹³C/¹²C)_standard] − 1 and VPDB is Vienna PeeDee Belemnite. Injection alone cannot account for the higher δ¹³C of CO₂ values at Olla; therefore, the integration of complementary datasets is required to evaluate additional in situ processes.

Owing to their inert nature, noble gases provide a powerful tool for tracing and quantifying physiochemical processes associated with CO₂ injection^{4,8,12,13,14,15,16,17,18,19}. The average air-corrected helium isotope (³He/⁴He) ratio relative to the atmospheric ratio (R/R_A, where R = ³He/⁴He_sample and R_A = ³He/⁴He_air) was determined for the Olla and Nebo-Hemphill oil fields to be 1.76 ± 0.31 R_A and 0.46 ± 0.12 R_A, respectively (Extended Data Fig. 1, Extended Data Table 2). The elevated He isotope values in the Olla Oil Field require an enhanced mantle-derived noble gases contribution, which is supported by additional noble gas data (Extended Data Fig. 1, Methods), associated with injected CO₂, which were sourced from the Black Lake Oil Field.

In addition to characterizing fluid origin, a combined noble gas and CO₂ isotope approach provides insight into processes associated with CO₂ trapping (for example, ref. ⁴) or methanogenesis following injection. ³He is inert and insoluble, with no significant sources within the crust. Thus, variations in CO₂/³He post-injection are directly attributable to the addition or removal of CO₂ within the system^4,20. Although two samples at Olla (samples O5 and O6) have CO₂/³He within the mantle range (2 ± 1 × 10⁹)_, consistent with the injection of mantle-derived fluids, the majority of the ratios are lower and indicate CO₂ removal (Fig. 2a, Extended Data Table 2). The difference between the highest CO₂/³He (O5, considered the most pristine sample) and the remaining samples provides a conservative estimate of post-injection CO₂ trapping/consumption of between 39% and 89%.

Fig. 2: The δ¹³C of CO₂ in the Olla (CO₂ injected field) samples.

a, The δ¹³C of CO₂ as a function of the CO₂/³He ratio. The dashed lines show endmember methanogenesis and dissolution (pH 7) fractionation trajectories. The tick marks represent the total amount of CO₂ trapping within the system, relative to sample O5. The shaded region represents trapping by the combination of both microbial methanogenesis and dissolution. The upper and lower methanogenesis:dissolution ratios (M:D) are 0.33 and 0.19, respectively, showing that dissolution accounts for approximately three times more CO₂ removal (M:D = 0.26) than microbial methanogenesis. The lines labelled ‘consumption’ show the portion of original injected CO₂ that has been removed by net microbial methanogenesis. b, The δ¹³C of CO₂ as a function of the δ¹³C of CH₄. The shaded region represents the CO₂ isotopic composition of the injectate from the Black Lake Oil Field into the Olla system. The Olla data are consistent with thermal re-equilibration with both reservoir temperatures (solid lines) and microbial methanogenesis (dotted lines). The 1σ level of uncertainty is within the symbol size.

A decrease in the CO₂/³He alongside increasing ⁴He and ²⁰Ne concentrations is observed within the Olla Oil Field (Extended Data Fig. 2, Methods). In the subsurface, ⁴He is produced via radiogenic decay of uranium and thorium and accumulates in formation water¹², which also contains atmosphere-derived ²⁰Ne. An anticorrelation between CO₂/³He and ⁴He and ²⁰Ne is observed in natural CCS analogues globally⁴, and indicates that the extent of water contact controls the magnitude of CO₂ trapping via dissolution or precipitation of CO₂ (refs. ^4,15). Within the observed pH range (7.0 ± 0.5) at Olla, dissolution will dominate over precipitation as a CO₂ trapping mechanism^4,10. However, as described above, the δ¹³C of CO₂ in the Olla gases is more enriched than can be explained by these processes alone (Fig. 2a).

Independent of the source of the injected CO₂, such positive δ¹³C values suggest significant modification from a starting composition that is unlikely to be heavier than 2‰ (refs. ^21,22). Notably, sorption to coal, hydrocarbon biodegradation to CO₂, and subsequent consumption and mixing processes could result in an increase in the δ¹³C of CO₂ (Methods). However, even from the most pristine sample (O5), it is unlikely that any of these processes could generate a fluid with the elevated concentrations of the isotopically enriched CO₂ observed at Olla.

Further constraints on the origins and processes affecting these gases are provided by the molecular-average and clumped isotope compositions of the associated CH₄. We observe a relatively high δ¹³C of CO₂ (13.6 ± 3.2‰) and CH₄ (−47.7 ± 1.7‰) at Olla compared with those observed at Nebo-Hemphill (3.5 ± 3.6‰ and −56.4 ± 3.6‰ for CO₂ and CH₄, respectively) (Fig. 2b, Extended Data Fig. 3a, Extended Data Table 2). The higher δ¹³C of CH₄ at Olla could be interpreted as evidence that the Olla Oil Field contains a higher proportion of thermogenic gas, or thermogenic gas of higher thermal maturity, whereas the Nebo-Hemphill Oil Field contains more microbial gas or thermogenic gas of lower thermal maturity. Using our multi-isotope approach, we present a model that is consistent with the observed higher δ¹³C values of CH₄ and an increase in the δ¹³C of residual CO₂. In this model, Olla is assumed to have had a pre-CO₂-EOR composition resembling Nebo-Hemphill, but following CO₂-EOR, microbial activity converted significant amounts of injected CO₂ to CH₄ (Extended Data Fig. 4). We arrive at this conclusion on the basis of the following observations.

First, we report a difference between the δ¹³C of CH₄ and the δ¹³C of CO₂ of 53.7–64.4‰ at the Olla Oil Field and of 50.8–68.1‰ at the Nebo-Hemphill Oil Field (Fig. 2b). The observed carbon isotope fractionations between co-existing CO₂ and CH₄ are consistent with thermodynamic isotopic equilibrium at temperatures between 44.6 °C and 76.3 °C at Olla and between 29.8 °C and 85.3 °C at Nebo-Hemphill²³, which overlap the present-day reservoir temperatures within these fields (29.2–50.7 °C and 29.7–57.1 °C, respectively), which suggests that the systems are approaching isotopic equilibrium under current reservoir conditions (Methods). The isotopic approach to equilibrium under reservoir conditions appears to be a result of microbial cycling of carbon (that is, methanogenesis and anaerobic oxidation of methane (AOM); evidence for AOM is apparent in the clumped isotopologues, see below). Equilibrium with reservoir conditions between the δ¹³C of CO₂ and the δ¹³C of CH₄ has previously been observed under similar geological conditions²⁴.

Second, we observe a range in the two measurable clumped isotopologues of CH₄, Δ¹³CH₃D and Δ¹²CH₂D₂, at Olla of 3.45–5.62‰ and 9.13–12.4‰ (Fig. 3, Extended Data Table 2), respectively, which correspond to apparent temperature ranges of ${29}_{-12}^{+14}$ °C to ${128}_{-16}^{+17}$ °C and ${87}_{-11}^{+13}$ °C to ${132}_{-23}^{+30}$ °C, respectively, and are approaching equilibrium with current reservoir conditions. Clumped isotope compositions provide an independent constraint on the origin of CH₄ in petroleum systems^{25,26,27,28,29,30}. Thermogenic CH₄ appears to dominantly form under internal isotopic equilibrium^26,30 whereas microbial CH₄ is highly variable in its clumped isotope compositions, with both non-equilibrium^28,29 and apparent equilibrium signatures^{24,26,28,29,31,32,33,34}. For Olla, we expect thermogenic-CH₄-generation temperatures to have exceeded 163 ± 18 °C, from independent maturity constraints provided by biomarkers³⁵. Two samples from Olla (O4 and O6) appear to be in internal isotopic equilibrium (about 125 °C and about 95 °C) but have lower apparent temperatures than expected for pure thermogenic CH₄. These temperatures are consistent with a component of microbial CH₄ formed at or close to equilibrium with current reservoir conditions. The remaining samples have Δ¹³CH₃D apparent temperatures of ${28.7}_{-12.3}^{+13.6}$ °C to ${74.0}_{-15.0}^{+16.8}$ °C, within error of the present-day reservoir temperature for these fluids (Methods), consistent with in situ microbial methanogenesis and AOM. Furthermore, these samples exhibit deficits in Δ¹²CH₂D₂, consistent with fluids that are dominated by microbial CH₄^28,29,34 in a system approaching equilibrium via active methanogenesis and AOM. In contrast, the Δ¹³CH₃D-based temperatures of ${77.9}_{-12.3}^{+13.5}$ °C to ${166}_{-29}^{+34}$ °C at Nebo-Hemphill suggest that these fluids are dominated by thermogenic CH₄, although, similarly to Olla, the Δ¹²CH₂D₂ clumped isotopologues also suggest a microbial CH₄ contribution.

Fig. 3: Δ¹²CH₂D₂ versus Δ¹³CH₃D of the measured Olla and Nebo-Hemphill samples.

The clumped isotopologue space illustrates whether measured CH₄ is at internal thermodynamic equilibrium (black line) or not. The thermodynamic equilibrium curve is calculated following ref. ⁴¹. The shaded cross represents thermal equilibration to current reservoir temperatures in each isotopologue. The arrows represent the theoretical trends for methanogenesis (dark grey) and AOM (light grey) ²⁹. The 1σ level of uncertainty is shown on the measured samples.

Third, we identify evidence for the biodegradation of hydrocarbonsin the molecular geochemistry of oils produced from both the Olla and the Nebo-Hemphill oil fields (Methods). In the case of Olla, we also report an elevated δ¹³C of propane (Extended Data Fig. 3b). The presence of oils co-existing with propane in these reservoirs preclude the possibility of high-temperature cracking as the cause of the positive δ¹³C values and suggests that the propane has been subjected to biodegradation³⁶, consistent with low-temperature microbial activity within the field.

Fourth, small subunit (SSU) ribosomal RNA gene sequencing of microbial communities identifies the presence of hydrogenotrophic methanogens, methanol and methylotrophic methanogens, and anaerobic methanotrophs (ANME) assigned sequences (Extended Data Fig. 5). These were sampled from the formation water in two wells within each field alongside previous microbiology analysis of these fields³⁷. Co-occurrence of both methanogenic and methanotrophic Archaea shows the potential for conversion of CO₂ to CH₄ and vice versa, consistent with isotopic signatures at intramolecular and intermolecular equilibrium. However, equilibration is probably rate limited by AOM, as the abundance of methanogens was 100 times higher than ANME, and AOM probably proceeds at a lower rate per cell.

Combined independent geochemical and microbiological data support our assertion that oil field microbes convert injected CO₂ to CH₄ at Olla. This study identifies the microbial processes associated with the injection of CO₂ into the deep subsurface operating on decadal timescales and the scale of the processing.

To quantify the impact of these microbial processes on the Olla Oil Field, we have constructed a model that considers both dissolution and microbial methanogenesis (Fig. 2a, Methods). From our most pristine sample (O5, highest CO₂/³He), we find that to match the isotopic composition of CO₂ with its corresponding CO₂/³He, dissolution must account for 3.1 ± 1.1 times more CO₂ removal than hydrogenotrophic methanogenesis. From this model, net microbial methanogenesis is estimated to have consumed a minimum of between 13% and 19% of the post-injection CO₂, depending on sample location, whereas dissolution is responsible for removal of as much as 74% more CO₂. This estimate is independently confirmed when CO₂ concentrations are compared with the δ¹³C of CO₂ values (Extended Data Fig. 6). On the basis of the elevated δ¹³C of CO₂, we recognize that our most pristine sample has probably also undergone a significant degree of modification, consistent with our estimate being conservative, and that CO₂ concentrations post-injection were likely to be higher and variable across the field.

Our combined noble gas and stable isotope approach allows estimates of in situ net microbial methanogenesis rates from within a natural system to be calculated. By extrapolating our results over the 29 years between the cessation of injection (1986) and sampling (2015), and assuming 13–19% microbial consumption of CO₂ since injection (from the hybrid methanogenesis–dissolution model, Fig. 2) with the remaining injected CO₂ volume, we calculate that a minimum of 1.15 × 10⁷–1.72 × 10⁷ m³ (STP) of microbial CH₄ has been produced at a minimum rate of 73–109 mmol CH₄ m⁻³ (STP) yr⁻¹. Previous estimates for CO₂ reduction following methanogenic oil degradation by hydrogenotrophic methanogens in lab microcosm incubations at similar temperatures are significantly lower (about 0.01–0.15 mmol CH₄ m⁻³ (STP) yr⁻¹)³⁸ by comparison.

This identification and quantification of CO₂-EOR-related microbial methanogenesis at Olla has significant implications, and warrants reconsideration of reservoir processes at similar sites worldwide. The fate of injected CO₂ in other oil fields (for example, Belridge Diatomite Formation, Lost Hills Oil Field, California, USA^17,39) has never been systematically explored using this integrated approach. Similarly to Olla, the Lost Hills Field has an elevated δ¹³C of dissolved inorganic carbon (DIC) (20.16–23.61‰)⁴⁰ compared with the injected δ¹³C of CO₂ (−30.1‰)³⁹. The fractionation between δ¹³C of DIC and CH₄ suggests that the Lost Hills system is approaching equilibrium with current reservoir conditions and the inverse correlation between the δ¹³C of the post-injection residual CO₂ and concentration is consistent with microbial methanogenesis. Furthermore, data from Nebo-Hemphill, as well as naturally CO₂-rich hydrocarbon systems in the Pannonian Basin, Hungary, are also consistent with significant microbial methanogenesis based on the CO₂/³He and the δ¹³C of CO₂ (Extended Data Fig. 7, Methods). These examples suggest that hydrogenotrophic methanogenesis probably occurs in both natural and injection-perturbed systems when reservoirs have favourable physiochemical conditions.

We conclude that although methanogenesis is not the dominant CO₂ sink, it can represent a substantial process within both natural and perturbed CO₂ fields and may be significantly more prevalent than previously considered. Even with the most conservative estimates of CO₂ conversion by methanogenesis in the Olla Oil Field, we find that, so far, as much as 13% to 19% of the emplaced CO₂ in sampled sections of the field has been consumed by methanogens. This is less than the amount of CO₂ that has been dissolved into water, but similarly occurs at significant rates on engineering timescales. CH₄ is less soluble and more mobile than CO₂ and therefore there is an enhanced risk of gas loss associated with CH₄ production due to microbial methanogenesis. Depleted hydrocarbon reservoirs have the second-largest CO₂ storage potential^1,3 after deep saline aquifers. If physicochemical and environmental conditions within these reservoirs (for example, low temperature, low salinity and high substrate availability) are conducive to microbial methanogenesis, we suggest that these microbial processes should be considered as criteria for future CCS site selection to ensure low-risk long-term storage. In addition, integrated studies like this may prove essential to effectively monitor CO₂ storage and the biogeochemical processes that result as a consequence of it.

Methods

Modelling

Here we provide extra details of the δ¹³C-CO₂ fractionation modelling of methanogenesis and dissolution. In our models (Fig. 2a, Extended Data Fig. 6), we use the sample with the highest measured CO₂/³He (most pristine sample, O5) as a reference point to calculate the change in CO₂/³He and δ¹³C-CO₂ with CO₂ processing (that is, methanogenesis and dissolution). We assume there is no ³He loss from the gas phase by dissolution or methanogenesis and therefore changes in the fraction of the CO₂ cause changes in the CO₂/³He.

Dissolution is modelled as open-system Rayleigh fractionation⁴.

$${\delta }^{13}{C}_{{\rm{C}}{\rm{O}}2}={\delta }^{13}{C}_{{\rm{C}}{\rm{O}}2i}\times {f}^{(\alpha -1)},$$

(2)

where f is the fraction of the original CO₂ remaining in the reservoir, δ¹³C_CO2 and δ¹³C_CO2i are the carbon isotopic composition of the CO₂ for post-dissolution and in the initial CO₂ respectively, and α is the fractionation factor. The fractionation factor for dissolution (α_d = 1.0038 ± 0.0012) was calculated based on the average measured pH¹⁰. Precipitation curves (Fig. 2a) can be modelled in the same way using fractionation factors (α_p = 1.0086 ± 0.0012) calculated for the current reservoir temperature²¹.

From consideration of the carbon isotopes, which appear to be approaching equilibrium under current reservoir conditions, net methanogenesis has been modelled as a closed-system equilibrium process (equation (2), Fig. 2a). The fractionation resulting from this cycling, α_m, is 0.9363 ± 0.003, which is consistent with thermal equilibration at current reservoir temperatures²³.

$${\delta }^{13}{C}_{{\rm{C}}{\rm{O}}2}={\delta }^{13}{C}_{{\rm{C}}{\rm{O}}2i}\mbox{--}({\alpha }_{m}-1)(1-f)\times \mathrm{1,000}$$

(3)

Within the Olla Oil Field, it is likely that there is a combination of both microbial cycling and dissolution. As a result, the net effect of microbial methanogenesis and dissolution can be described using equation (3).

$${\delta }^{13}{C}_{{\rm{C}}{\rm{O}}2}={\delta }^{13}{C}_{{\rm{C}}{\rm{O}}2i}\times {f}^{(\alpha -1)}\mbox{--}({\alpha }_{m}-1)(1-f)\times \mathrm{1,000}$$

(4)

where f is the fraction of CO₂ remaining after trapping/consumption and α_A is the apparent fractionation factor for the system. α_A depends on the degree of dissolution compared with methanogenesis and is described as⁴²:

$${\alpha }_{A}={F}_{d}{\alpha }_{d}+(1-{F}_{d}){\alpha }_{m}$$

(5)

where F_d is the fraction of CO₂ of the total CO₂ trapped/consumed due to dissolution.

Sample collection and analysis

Samples for bulk gas composition, stable isotope, noble gas and clumped isotope analysis were collected directly from the wellhead using standard techniques following methods from refs. ^8,17,18,24. Noble gas isotope determination was conducted in the Noble Laboratory at the University of Oxford, where the analysis of hydrocarbon gases is well established^{8,17,18,43,44,45}. For bulk composition and C and H stable isotope analysis, gas cylinders were shipped to Isotech in Champaign, Illinois, USA. The standard procedures for these are described in ref. ⁴⁶. CH₄ clumped isotopologues were determined at the California Institute of Technology following the procedures described in ref. ⁴⁶.

Microbiological analysis was conducted at an internal ExxonMobil Research and Engineering Company Facility. The produced water samples were collected and filtered immediately from wells O1, O4, NHH3 and NHH6. Water was passed through 0.22-μm SVGPL10 RC Sterivex GP filters (EMD Milipore) until filters clogged in the following amounts: O1, 1,350 ml; O4, 1,350 ml; NHH3, 1,750 ml; NHH4, 300 ml. Filters were stored on ice immediately and stored at −80 °C until DNA extraction. DNA was extracted from the Sterivex filters using the DNeasy PowerWater Sterivex Kit (Qiagen). The SSU rRNA gene was amplified with V4V5 specific primers (https://vamps2.mbl.edu/) targeting Bacteria. The SSU rRNA gene was amplified with specific primers targeting Bacteria: forward primers (967F) CTAACCGANGAACCTYACC, CNACGCGAAGAACCTTANC, CAACGCGMARAACCTTACC and ATACGCGARGAACCTTACC; reverse primer (1064R) CGACRRCCATGCANCACCT. A second set of amplifications was performed using specific primers targeting Archaea: (Arch2A519F) CAGCMGCCGCGGTAA and (Arch1017R) GGCCATGCACCWCCTCTC. Illumina MiSeq 2x300bp paired-end sequencing was performed by Mr. DNA. Microbiome bioinformatics were performed with QIIME 2 2017.4⁴⁷. Raw sequence data were demultiplexed and quality filtered using the q2-demux plugin followed by denoising with DADA2⁴⁸. Amplicon sequence variants were aligned with mafft and used to construct a phylogeny with fasttree2^49,50. Taxonomy was assigned to amplicon sequence variants using the q2‐feature‐classifier⁵¹ classify‐sklearn Naïve Bayes taxonomy classifier against the Greengenes 13_8 99% operational taxonomic unit (OTU) reference sequences⁵².

Geological and production history

The Olla Oil Field is located in La Salle Parish, Louisiana, in the northern Gulf of Mexico and was CO₂ injected between 1983 and 1986 for enhanced oil recovery (CO₂-EOR). The Nebo-Hemphill Oil Field lies 20 km to the southeast of the Olla Oil Field and provides a control field for this study, having had no CO₂ injected (Fig. 1). Both fields are producing from the Palaeocene–Eocene Wilcox Group, which in northern Louisiana is composed of 600–1,500 m of shallow marine clastic sediments to fluvial deltaic deposits and up to 6-m-thick coal beds^53,54. The group is bound either side by the Palaeocene Midway Group and the Eocene Clairborne Group confining units⁵⁴. Typically, the Wilcox Group is subdivided into the lower (sandstone with coal beds), middle (sandstone, lignite and clay) and upper (deltaic deposits) intervals, with the ‘Big Shale’ informal unit separating the middle and upper groups^53,55. Late Cretaceous intrusive igneous bodies in northern Louisiana, west-central Mississippi and southeastern Arkansas are thought to have affected the regional heat flow, which may have caused an increase in the thermal maturity of the Wilcox Group during the early Palaeocene^56,57.

Gas/oil ratios are variable within the field, and do not seem to be associated with proximity to the known gas cap and higher gas/oil ratios may instead reflect gas coming out of solution as bottom hole pressure decreases.

CO₂ is present in low concentrations (1.02 ± 0.69 mol%) within the Nebo-Hemphill Oil Field, which suggests that there is some natural background CO₂ present in both fields. Between March 1983 and April 1986, the 2,800-ft sandstone (within the middle Wilcox Group) in the Olla Oil Field underwent a pilot project for CO₂ flooding. The CO₂ injected into the field was immiscible and injected in the gas phase (that is, not supercritical). It is assumed that a negligible amount of CH₄ was injected alongside the CO₂. Previous studies on the Olla Oil Field have assumed that all the injected CO₂ was retained within the 2,800-ft sandstone^9,10. However, the initial CO₂ injection report⁷ states that “mole percentages in excess of 10% more than likely indicate production of injected CO₂”. Only 1 sample outside of the 2,800-ft sandstone has CO₂ concentrations less than 10% (that is, all other CO₂ concentrations are greater than 10%). Notably, concentrations greater than 10% in other producing zones have previously been reported¹⁰. These signatures are consistent with injected CO₂ migration across the different reservoirs and remaining within the formations upon sampling. If such high concentrations of CO₂ were present in this field before injection, CO₂ would not have been extracted and transported from the Black Lake Oil Field and would have instead have been extracted from the producing fluids of CO₂ rich fluids at Olla. In addition, we see a clear mantle-derived noble gas signature (see below), which is most evident in the He and neon (Ne) isotopes (Extended Data Fig. 1), in the Olla Oil Field, which shows a clear incursion of injection CO₂ into the proximal lithologies.

A total of 2.2 × 10⁸ m³ of CO₂ was injected into the Olla Oil Field and was sourced from the nearby Black Lake Oil Field (produced from the Lower Cretaceous carbonates of the Petit Formation) in Natchitoches Parish, Louisiana (Fig. 1). Approximately one-third of the injected CO₂ remained within the reservoir after the cessation of the EOR project⁷. It is important to note that given the age of this CO₂-EOR project, we were unable to obtain constraints such as the ratio of injected CO₂ to reservoir pore volume that are sometimes cited to quantify the scale of the EOR project and the distribution it achieves within the subsurface. Further, we anticipate that the injected CO₂ is not uniformly distributed within the pore volume of the Olla Oil Field due to different phases of EOR and gas cycling.

In 1984, water flooding also started within the field. In contrast, Nebo-Hemphill has never undergone any type of EOR.

Molecular geochemistry

Whole-oil gas chromatography (WOGC) was performed to provide a hydrocarbon fingerprint of each oil, and open-column liquid chromatography was utilized to separate and quantify the saturate, aromatic, resins and asphaltene fractions of each oil. In general, we see similar fluid compositions between the two fields. The WOGC shows evidence of marine-sourced oils with a smooth distribution of n-alkanes, pristine/phytane ratios of approximately 2 and low wax content. The notable differences are probably due to varying levels of biodegradation illustrated by the loss of n-alkanes in some samples and American Petroleum Institute (API) gravity ranging from 21 to 42. The similarities in the geochemical signatures of these oils provide a basis to use the Nebo-Hemphill Oil Field as a control site for any biogeochemical differences that may arise as a result of the injection of CO₂ at Olla.

Source facies

Gas chromatography mass spectroscopy (GC/MS) was utilized to analyse the saturate and aromatic fractions to provide the biomarker distributions of each oil. The interpretation of a dominantly marine source is based on the following evidence: (1) a smooth distribution of tricyclic terpanes with C₂₃ being the maximum along with extended hopanes (C₃₁–C₃₅) decreasing in order is a classic marine source signature; (2) the high presence of C₃₀ hopane compared with C₂₉ hopane is indicative of a marine clastic source; (3) a higher abundance of C₂₈ bisnorhopane compared with oleanane suggest a higher marine input compared with the above source⁵⁸; (4) a greater abundance of C₂₆ tricyclic terpane as compared to C₂₅ tricyclic terpane provides evidence of some terrigenous input. Consistent with previous interpretations, we consider this source to be a part of the Wilcox Group oils, more specifically Wilcox Group I as originally described by refs. ^59,60.

Thermal maturity

We adopted standard saturate and aromatic biomarker ratios to develop apparent thermal maturity estimates for each of the oil samples analysed. We discuss these in the context of apparent thermal maturities because we recognize these estimates are based on the bulk weighted average composition of the fluid. The interpretation of the GC/MS analyses of the 12 oils yielded a thermal maturity range of R_o equivalent of 0.9–1.25%, which is classified as main-to-late-stage oil. The interpretation of this thermal maturity range is based on the following: (1) the presence of higher concentrations of high-molecular-weight compared with low-molecular-weight triaromatic steriods, described as % ΔTAS (20–57%); (2) the isomerization of C₂₉ ααα−20S relative to C₂₉ ααα−20R steranes (0.45–0.48)⁶¹; (3) the isomerization of 1,2,5-trimethylnapthalene to 1,3,7-trimethylnapthalene represented as ratio TMNr (0.48–0.67) as defined by ref. ⁶²; (4) the ratio of C₂₃ tricyclic terpane to C₃₀ hopane, which is calculated as T23/(T23 + H30) (0.10–0.37), where the increased presence of C₂₃ tricyclic represents higher maturity.

Alteration

The use of WOGC and GC/MS analyses help provide a picture of the alteration history of these oils. The shallow Nebo-Hemphill NHH3 fluid is the most biodegraded end member of the samples tested, based on the apparent lack of n-alkanes, an API gravity of 21 and the presence of demethylated hopanes. This sample would be classified as severely biodegraded based on the classification of ref. ⁶³. The remaining Nebo-Hemphill samples have APIs of 32–41 and are classified as unbiodegraded to very slight biodegradation⁶³. The Olla Oil Field fluids show evidence of reduced light n-alkanes, display similar n-alkane-to-isoprenoid ratios and have APIs of 33 to 37, suggesting similar levels of biodegradation between the fluids. Therefore, the Olla fluids would be classified as having very slight biodegradation⁶³.

Summary of previous findings

The formation water geochemistry, molecular oil composition and produced gas composition (including δ¹³C) in the Olla and Nebo-Hemphill oil fields have been previously characterized^9,10,11. Microbial methanogenesis was first hypothesised for the Olla Oil Field based on its enriched δ¹³C of CO₂ (ref. ⁹). However, owing to the differences in the isotopic composition between the 2,800-ft sandstone and the other producing formations with the injected material, they concluded that the injected CO₂ was not present in the producing zones¹⁰ and consequently any microbial methanogenesis was independent of injection and is probably associated with microbial oxidation of hydrocarbons. The enhanced methanogenesis within the Olla Oil Field compared with other fields produced from the Wilcox Group was thought to be due to the ideal geochemical conditions (that is, suitable range of pH, temperature and salinity, and lack of alternative electron acceptors)¹⁰. They conclude that the injected CO₂ was most likely dissolved and trapped in a gas phase¹⁰. Further investigation of microbial methanogenesis found that there was little correlation between crude oil biodegradation and methanogenesis; however, the extent of oil biodegradation was correlated with the temperature, salinity and depth of samples¹¹. In addition, there has also been an investigation of the microbial community composition within the Olla reservoir³⁷, which found methanogens present within the system, and no significant difference in microbial communities between the 2,800-ft sandstone and other producing zones³⁷. With the addition of the inert noble gas isotopes and clumped isotopes, we are able to re-investigate the fate of CO₂ within the Olla reservoirs.

Formation water chemistry in the Olla Oil Field

The pH and alkalinity of the formation waters within the Olla Oil Field are between 6.5 and 7.5 and between 16.3 meq kg⁻¹ and 57.7 meq kg⁻¹, respectively¹⁰. The most dominant anion was Cl⁻ (1,122–1,621 mmol l⁻¹) and the most dominant cation measured was Na²⁺ (1,108–1,630 mmol l⁻¹) (Extended Data Table 1), which are typical of basinal brines⁶⁴ and in agreement with previous measurements¹⁰. Other cations present include: Mg²⁺ (119–329 ppm), Ca²⁺ (80–500 ppm), K⁺ (223–524 ppm), Ba²⁺ (95–165 ppm), Si⁴⁺ (10.1–12.5 ppm) and Sr²⁺ (77–215 ppm). Measured iron concentrations are less than 0.1 ppm. With the exception of sample O1 where SO₄²⁻ concentrations are 62 ppm, SO₄²⁻ concentrations are less than 2 ppm. Both the iron and SO₄²⁻ concentrations are in agreement with those previously reported¹⁰.

Identification of injected material

CO₂ concentrations in the injected Olla Oil Field (18 ± 12mol%) are significantly greater than the Nebo-Hemphill Oil Field (1.02 ± 0.69 mol%). CH₄ concentrations at Olla are 77 ± 13 mol% (STP) compared with 92.6 ± 8.2 mol% (STP) at Nebo-Hemphill. However, the δ¹³C of CO₂ of the injected material (0.85 ± 0.91‰) is isotopically distinct from the δ¹³C of CO₂ now within Olla (13.6 ± 3.2‰) (Fig. 2b). The injected CO₂ was sourced from the Black Lake Oil Field, which is located close to the Sabine Island Complex. The Sabine Island Complex forms part of the Sabine Uplift basement high and has been suggested to be a conduit for mantle-derived fluids⁸. However, the measured δ¹³C of CO₂ values at Black Lake (0.85 ± 0.91‰)¹⁰ are elevated relative to mantle CO₂ (−2‰ to −5‰; refs. ^21,22), probably reflecting some degree of biological modification when the reservoir was shallower and at temperatures favourable to microbial activity. An alternative interpretation for the elevated δ¹³C of CO₂ at Black Lake is that it is a mixture of mantle-derived and crustal-derived CO₂; however, this would increase the CO₂/³He above that expected in the mantle, which is not observed in the Olla samples.

As the stable isotopes of carbon are frequently modified by chemical or biological processes, the inert nature of noble gases means that they are ideal tracers for injected CO₂ in such scenarios. As injectate samples were not collected during CO₂-EOR, we do not have access to the noble gas compositions of the material injected into the Olla Oil Field. Nevertheless, it is reasonable that the noble gases record pre-injection conditions at the nearby Nebo-Hemphill Oil Field, and that these are subsequently modified at the Olla Oil Field through mixing of the injected material during the CO₂ injection.

The ⁴He/²⁰Ne ratio is between 3,210 and 20,900 in both fields, which is significantly greater than the atmospheric value of 0.32 and thus there is no significant atmospheric contribution to these samples. The air-corrected He isotopic ratio (³He/⁴He), reported relative to the atmospheric ratio (R_C/R_A where air is 1R_A), in the Olla Oil Field is 1.84 ± 0.10R_A and 0.46 ± 0.08R_A at Nebo-Hemphill (Extended Data Table 2). These are both higher than a typical hydrocarbon system where the majority of He is derived from crustal radiogenic production (³He/⁴He  = 0.02R_A; ref. ¹²), and higher than other fields within the Gulf of Mexico region^8,18,43. The elevated ³He/⁴He values are the result of a resolvable mantle contribution. Using a simple two endmember mixing model, with subcontinental lithospheric mantle (³He/⁴He = 6.1; ref. ⁶⁵) and crustal endmembers, we resolve an average 7.8% mantle contribution of mantle He in Nebo-Hemphill Oil Field. It is likely that the mantle fluid contribution is coming from the Little Creek collapse structure, which is situated between the two fields⁶⁶. At Olla, the average mantle contribution to ³He is 30%, which suggests that there is an additional source of mantle fluids to this field, which is reasonably explained by the injectate sourced from the Black Lake Oil Field. Ne isotope ratios are elevated at Olla (²⁰Ne/²²Ne 10.3 ± 0.2 and ²¹Ne/²²Ne 0.0399 ± 0.002) compared with Nebo-Hemphill (9.85 ± 0.09 and 0.0335 ± 0.001, respectively) and air (9.8 and 0.029) (Extended Data Fig. 1, Extended Data Table 3)^67,68,69, again suggesting that the injected material contains mantle-derived fluids.

The ⁴⁰Ar/³⁶Ar values at Nebo-Hemphill are significantly greater than the atmospheric value (298.56; ref. ⁷⁰), with measured ratios of 370 ± 3 to 604 ± 5. The Olla Oil Field has a higher ⁴⁰Ar/³⁶Ar (1,476 ± 13 to 2,766 ± 24) (Extended Data Table 3). Elevated ⁴⁰Ar/³⁶Ar ratios can be a result of both an excess of radiogenic ⁴⁰Ar and mantle-derived fluids^12,71. The difference in the ⁴⁰Ar/³⁶Ar ratios could be due to variable amounts of radiogenic production or fundamental differences in the extent of contact with formation waters between the two fields. However, as all the wells sampled are within the same geological formation, the two fields are in relatively close proximity and they have indistinguishable ⁴He concentrations, it seems unlikely that this would be the cause. Thus, the difference is probably a result of the injected CO₂, which probably had mantle-derived ⁴⁰Ar/³⁶Ar excesses, concordant with the observed mantle-derived He and Ne isotopes. Finally, whereas xenon (Xe) isotope ratios at Nebo-Hemphill are indistinguishable from air, the Olla Oil Field showed excesses in some Xe isotope ratios relative to air. For example, the ¹²⁹Xe/¹³⁰Xe ratio is 6.59 ± 0.01 compared with an atmospheric value of 6.49 (Extended Data Table 3); this excess in ¹²⁹Xe relative to ¹³⁰Xe is consistent with a contribution from mantle fluids with an end member ¹²⁹Xe/¹³⁰Xe value of 8.2 (ref. ⁷¹). The Olla Oil Field therefore shows a clear mantle component that is not present within the Nebo-Hemphill control field.

The detection of injected fluids within the Olla Oil Field using the conservative noble gas tracers differs from the findings of the previous studies that used the δ¹³C of CO₂ to conclude that injected CO₂ is no longer or never was present in the majority of the sampled wells (for example, ref. ¹⁰). The subsurface behaviour of the injected CO₂ within the system can therefore be re-examined using a combination of noble gases and stable isotopes.

Dissolution within the Olla Oil Field

CO₂ is highly soluble in water under reservoir conditions. The most likely trapping mechanisms for free-phase subsurface CO₂ are dissolution or precipitation as carbonate minerals mediated by dissolution^4,21. A decrease in the CO₂/³He correlates with increasing ⁴He and ²⁰Ne concentrations within the Olla Oil Field (Extended Data Fig. 2). ²⁰Ne (and often ⁴He)⁴ are derived from the formation water, suggesting that CO₂ trapping in the Olla Oil Field is also proportional to the amount of formation water that has been contacted and degassed, consistent with dissolution within the field⁴. Dissolution (solubility trapping) has previously been found to be a major sink for CO₂ (ref. ⁴) and imparts a slight pH-dependent fractionation towards a light isotopic composition²¹. However, this fractionation effect is marginal compared with the amount of CO₂ that can be trapped and the fractionation that can be imparted by methanogenesis–AOM cycling (Fig. 2a).

Given the presence of discontinuous coal beds within the Wilcox Group^57,72, it is possible that some of the injected CO₂ may have been sorbed onto these coal beds as CO₂ has a much greater affinity to coal than CH₄ (ref. ⁹). The resulting isotopic fractionation imparted by this process remains poorly constrained, with some studies suggesting a greater than 5‰ shift^73,74 whereas others suggest negligible effects^75,76,77 on the δ¹³C of CO₂. Coal sorption would cause fractionation in the opposition direction of the CH₄ isotopic signature. Given the heavier δ¹³C of CO₂ and δ¹³C of CH₄ isotopic signatures in the injected Olla Oil Field, it is unlikely that CO₂ sorption to coal is a major CO₂ sink in the field, in agreement with previous findings¹⁰.

Subsurface behaviour of carbon at Olla

The Olla samples are enriched in both the δ¹³C of CO₂ and the δ¹³C of CH₄ compared with Nebo-Hemphill and the injected CO₂ values. The offset between the δ¹³C of CH₄ and the δ¹³C of CO₂ in the Olla Oil Field are close to those expected under thermally driven isotope exchange equilibrium for the observed temperatures (Fig. 2b). Carbon isotopes of CO₂ and CH₄ approaching equilibrium in hydrocarbon systems with current reservoir conditions have previously been observed²⁴. However, the thermal equilibrium of CO₂ and CH₄ is kinetically limited at the current temperatures (29.2–50.7 °C) in the Olla reservoirs and would not occur on the timescale of injection, without a process to enhance the reaction rate⁷⁸. A combination of both microbial methanogenesis and AOM active within the field would allow for the system to be approaching equilibrium under the current thermal conditions within the reservoir. Further observations detailed below provide evidence for this conclusion and allow for the net microbial methanogenesis to be quantified.

The clumped isotopologues of CH₄ (Δ¹³CH₃D and Δ¹²CH₂D₂) can provide independent constraints on its origin within petroleum systems^{25,26,27,28,29}. Thermogenic CH₄ is expected to have a composition consistent with the source-rock maturation temperature, which is typically at or above 157 °C (ref. ²⁵). At Olla, we expect CH₄ generation to have occurred at 163 ± 18 °C (based on biomarker maturity estimates following ref. ³⁵). Microbial CH₄ is variable in its clumped isotopic compositions with both non-equilibrium^28,29 and apparent equilibrium signatures reported in samples obtained from the deep subsurface^26,29. Microbial methanogenesis forms CH₄ characterized by low Δ¹³CH₃D and extremely low Δ¹²CH₂D₂ (refs. ^28,29), whereas AOM tends to drive the isotopologue composition of CH₄ towards equilibrium, with equilibrium expected to be reached at a faster rate in Δ¹³CH₃D compared with in Δ¹²CH₂D₂ (refs. ^{29,30,34,79,80,81}). In the Olla Oil Field, we see Δ¹³CH₃D of 3.45–5.62‰ (apparent temperature of ${29}_{-12}^{+14}$ °C to ${128}_{-16}^{+17}$ °C) and Δ¹²CH₂D₂ of 9.13–12.4‰ (apparent temperature of ${87}_{-11}^{+13}$–${132}_{-23}^{+30}$ °C) (Fig. 3, Extended Data Table 2); these temperatures preclude a thermogenic origin for the CH₄ and are instead consistent with a microbial contribution. Samples O4 and O6 are in internal clumped isotopologue equilibrium at temperatures lower than expected for the thermogenic CH₄ and appear to be approaching equilibrium with current reservoir conditions. The remaining samples from Olla have already reached apparent equilibrium with the current reservoir conditions in Δ¹³CH₃D but have yet to reach apparent equilibrium in Δ¹²CH₂D₂ probably due to kinetic effects. The clumped isotopologues at Olla are approaching equilibrium at reservoir temperatures, which would be accounted for by active AOM within the Olla Oil Field. This supports the conclusion that there is active microbial methanogenesis and AOM driving the carbon isotopes to approach equilibrium under current thermal conditions in the Olla reservoirs.

There is microbial degradation of the higher alkane gases (Extended Data Fig. 3) within the Olla Oil Field. Propane is preferentially consumed during microbial degradation and thus in biodegraded gases, propane will see the most pronounced δ¹³C enrichment³⁶. This δ¹³C enrichment is clear within the Olla gases (Extended Data Table 3) and supports active microbial hydrocarbon degradation within the field.

The presence of methanogenic activity and the idea of a methanogenic ‘hotspot’ in the Olla reservoir has been previously suggested^9,10. The environmental conditions for which methanogenesis can occur are relatively well defined. Within nutrient-poor environments such as oil reservoirs or coal beds, methanogens can survive at temperatures up to 80 °C, with the largest growth rates and thus greatest amount of methanogenic activity between 40 °C and 50 °C (refs. ^36,82,83,84). Methanogenic activity is greatest at near neutral pH and is inhibited at pH <4 and >9 (ref. ⁸⁴). Formation water salinity can also inhibit methanogenesis above Cl⁻ concentrations of 1,500 mM, with greatest growth between 200 nM and 600 mM (refs. ^82,83,85). The presence of ferric iron and sulfate at concentrations greater than 1 mM can also limit methanogenesis (as a result of the dominance of other bacteria)^86,87,88,89. Notably, the Olla Oil Field exhibits ideal environmental conditions for methanogenesis, in terms of pH (6.5–7.5)¹⁰, temperature (29.2–50.7 °C), salinity (1,094–1,622 mM), sulfate (<0.6 mM l⁻¹)¹⁰ and iron concentrations (≤0.1 mM l⁻¹) (Methods).

In addition, geochemical and isotopic parameters associated with microbial methanogenesis in organic-rich settings are also well established^{10,11,83,88,90}. Traditionally, carbon isotopes of CH₄ have been used to determine the presence of biogenic CH₄ (typically less than −60‰)^91,92,93(Extended Data Fig. 3a). However, methanogenesis can result in heavy biogenic CH₄ as the residual CO₂ pool gets increasingly elevated, consistent with observations at Olla (Fig. 2b). In addition, where methanogenesis has occurred, formation water alkalinity is greater than 10 meq kg⁻¹, the acetate concentrations are lower than 1 mM, the Ca/Mg ratios are less than 1.5 and the δ¹³C-DIC is greater than +20‰ (refs. ^87,88,90). Within the Olla Oil Field, the alkalinity ranges between 21.4 meq kg⁻¹ and 57.7 meq kg⁻¹, the average Ca/Mg ratio is 0.2 and the δ¹³C-DIC is 14.2–28.7‰, again indicating that methanogenesis has occurred within the field.

SSU rRNA gene-sequencing analysis of the microbial communities within the Olla Oil Field (samples O1 and O4) indicate that within the archaeal sequences, hydrogenotrophic methanogens, methanol and methylotrophic methanogens, and ANME assigned sequences are present (Extended Data Fig. 5). This microbial array would facilitate both microbial methanogenesis and AOM within the field.

During the cycling of carbon between methanogenesis and AOM, we would not expect a change in the isotopic composition of hydrogen. It is possible that as a result of microbial activity or the reaction kinetics that extracellular hydrogen is added to the system. This hydrogen could be sourced from the water and free hydrogen from the degradation of the higher hydrocarbon gases or from the oil. As a result, we are unable to determine whether the hydrogen isotopes within the system are also approaching equilibrium under current reservoir temperatures.

It is possible that the CO₂ and CH₄ isotopes in the Olla system are reflecting purely the kinetic process of methanogenesis. The difference in the δ¹³C of CO₂ and the δ¹³C of CH₄ at Olla (53.7–64.4‰) is consistent with that expected from microbial methanogenesis (about 58‰; ref. ⁹⁴). In such a scenario, all the CH₄ at Olla would be microbial in origin, thus requiring any thermogenic CH₄ component originally in the system to be overwhelmed. The high concentration and mixed thermogenic–microbial composition of CH₄ within the Nebo-Hemphill Oil Field (Fig. 3, Methods) is inconsistent with such a scenario. In addition, the clumped isotopologues at Olla show a clear need for AOM within the field and 16S rRNA gene sequencing shows active methanotrophs, which is not compatible with a scenario of purely microbial methanogenesis dominating the system. Therefore, future studies investigating the processes affecting CO₂ in potential CCS storage targets will benefit from a complete geological, geochemical and microbiological characterization of the field study site.

Subsurface behaviour of carbon in the Nebo-Hemphill Oil Field

The carbon isotopic composition of CO₂ and CH₄ at Nebo-Hemphill is depleted compared with the Olla Oil Field (Fig. 2b). Despite this, the difference in the δ¹³C of CO₂ and the δ¹³C of CH₄ at Nebo-Hemphill overlaps with that of Olla (50.8–68.1‰ and 53.7–64.4‰, respectively). The degree of isotopic fractionation at Nebo-Hemphill is consistent with thermal equilibrium at 29.8–85.3 °C, in agreement with present-day reservoir temperatures (29.7–57.1 °C). In addition, the δ¹³C of CO₂ at Nebo-Hemphill is enriched compared with that typically expected for crustal or mantle-sourced CO₂ (less than 2‰)^21,22, consistent with a small amount of microbial methanogenesis within the field. We propose that there is also microbial methanogenesis (in agreement with previous studies¹⁰) and AOM within the Nebo-Hemphill Oil Field; however, unlike at Olla, this has not been enhanced by any injection to the field.

Microbiological analysis of NHH3 and NHH6 show that hydrogenotrophic methanogens, methanol and methylotrophic methanogens, and ANME assigned sequences are present within Nebo-Hemphill, in agreement with both active methanogenesis and AOM within the field (Extended Data Fig. 5). Clumped isotopologues of CH₄ within the Nebo-Hemphill Oil Field give apparent temperatures for Δ¹³CH₃D of ${78}_{-12}^{+14}$–${166}_{-29}^{+34}$ °C and apparent temperatures for Δ¹²CH₂D₂ that are greater than expected for thermogenic CH₄ in the field (163 ± 18 °C, Fig. 3). They suggest that the CH₄ is likely to be dominantly thermogenic with some microbial contributions. Samples NHH1, NHH2 and NHH8 have apparent temperatures below that expected for thermogenic CH₄ and appear to be approaching low-temperature equilibrium in Δ¹³CH₃D, consistent with active methanogenesis and AOM cycling within the field.

Modelling of the Pannonian Basin

The unequivocal identification and quantification of CO₂-related microbial methanogenesis at Olla has globally significant implications, and warrants the reconsideration of reservoir processes at other sites worldwide. Re-examination of previously published datasets allows the elucidation of the significance of methanogenesis in different reservoir environments, especially in systems that have or had naturally high CO₂ concentrations, such as the Pannonian Basin. A significant proportion of the CO₂ found in natural gas fields within the Pannonian Basin is of mantle origin²⁰. The initial, CO₂/³He in the Szegholm North, Szegholm South, Ebes and Hajduszoboszlo gas fields was estimated to be 7.9 ± 5.4 × 10⁹, consistent with the European subcontinental lithospheric mantle range (0.6–40 × 10⁹) and the mid-ocean ridge basalt mantle range (2 ± 1 × 10⁹)^95,96,97 (Extended Data Fig. 7a). Since emplacement, significant CO₂ trapping in many fields was identified from lower observed CO₂/³He ratios (for example, the Szegholm North and South, Ebes and Hajduszoboszlo gas fields) and speculation that some of the CO₂ may have been converted into CH₄ (ref. ²⁰), although no viable mechanism was identified.

Using an average geothermal gradient of 45 °C km⁻¹ the temperature within these fields is predicted to be between 40 °C and 105 °C (with temperatures being up to 20 °C cooler at the time of CO₂ emplacement)⁹⁸, and thus within the range of microbially driven methanogenesis. In addition, the presence of thermogenic hydrocarbons would also ensure H₂ availability, similar to the Olla and Nebo-Hemphill oil fields. From our work in Olla, we surmise that the Pannonian Basin may be analogous to Olla and that hydrogenotrophic microbial methanogenesis active within the field may provide the missing mechanism for CO₂ removal, which was previously unidentified, and point to this mechanism as an important subsurface process operating on systems naturally enriched in CO₂ as well as anthropogenically CO₂ injected oil and gas fields.

In the absence of supporting clumped isotopologue data, if we assume that the system is methanogenesis driven we can nevertheless construct similar methanogenesis–dissolution models using published data from the Szegholm South Gas Field (where multiple CO₂ isotope data are available). By comparing the observed CO₂/³He ratios against the CO₂ concentration, a clear correlation emerges (Extended Data Fig. 7a). Attributing this to the effect of dissolution (a pre-requisite for microbial methanogenesis), it is possible to extrapolate back to an initial CO₂/³He end member value of 13.3 × 10⁹, consistent with previous estimates.

Our model assumes mantle-derived CO₂ in the field with a typical isotopic composition of δ¹³C of CO₂ = −5‰ (Extended Data Fig. 7b). Taking this as a starting point, 16.4 ± 0.8 times more dissolution than methanogenesis is required in the Szegholm South Gas Field to produce the measured δ¹³C of CO₂. As such, proportionally less methanogenesis, relative to the Olla Oil Field, would be expected here as the gas fields are not being constantly perturbed via water injection and therefore nutrient availability is probably much more limited. The presence of biogenic CH₄ (through CO₂ conversion) can also be clearly identified from the measured C₁/C_N in the Szegholm South Gas Field (0.946 ± 0.004), which is greater than what would be expected in a purely thermogenic system (0.909). Using the excess in the C₁/C_N ratios, we determine that 41 ± 5% of the CH₄ currently present in the Szegholm South Gas Field part of the basin is biogenic. Similar C₁/C_N excesses are seen in the other gas fields in the Pannonian Basin⁹⁹. In addition, other fields such as Szegholm North^20,99, Ebes^20,99, Hajduszoboszlo^20,99, Kismarja^20,99, Répcelak¹⁰⁰ and Mihàly¹⁰⁰ (Pannonian Basin), Subei¹⁰¹ (China) and JM Brown Bassett⁹⁵ (Permian Basin) also show geochemical signatures consistent with the occurrence of methanogenesis. Thus, we conclude that although methanogenesis is not the dominant CO₂ loss mechanism, it may play a substantial role that has not previously been considered in many natural systems.