Introduction

Carbon capture and storage (CCS) is a candidate technology to reduce the emission of CO2 from factories and power plants1,2. The underlying idea is to capture CO2 rather than emitting it into the atmosphere and to sequester it in deep subsurface formations such as deep saline aquifers or depleted oil and gas reservoirs3. This technology has the potential to reduce future global CO2 emissions by 20%2,4. The economical and technological feasibilities of CCS have been extensively studied over the past decade, mainly from a geological perspective5,6,7,8. Recently, microbial effects associated with CO2 storage also have been investigated in deep saline aquifers9,10, and the influence of CO2 injection on microbial community structure has been reported. A study revealed that CO2 injection changed the constituent of methanogens and sulfate reducers in the deep subsurface microbial community10. Oil reservoirs are also unique subsurface environments that are potential repositories for CO2 sequestration11,12 and often harbor active methane-producing microbial communities13,14,15. However, very little is known about the effect of CO2 sequestration on microbial communities and their functional role in deep subsurface oil reservoirs.

In the present study, we aimed to reveal how the microbial community and functions associated with methanogenesis respond to the increase in CO2 concentration in the oil reservoirs. We performed laboratory experiments that simulated field conditions from where samples were collected. Based on stable isotope tracer analysis, molecular analysis and thermodynamic calculations, we report the possibility that drastic changes of microbial communities and methanogenic functions can be induced by CO2 injection into deep subsurface high-temperature oil reservoirs.

Results

Effect of CO2 on methanogenesis in oil field microcosms

We selected a high-temperature oil reservoir, Yabase oil field, Japan, where we previously demonstrated the occurrence of in situ methanogenic activity15. Acetate, which is highly abundant in this reservoir (6–9 mM), is a primary precursor to methane. We previously found that methane was produced from acetate via syntrophic acetate oxidation coupled with hydrogenotrophic methanogenesis in the reservoir15. To determine if methanogenesis would continue if CO2 concentrations increased in subsurface oil reservoirs, we set up high-temperature and high-pressure incubation apparatuses (working volume: 1 l; Supplementary Fig. S1), mimicking the in situ oil reservoir (1,000–1,300 m deep, 53–65 °C, 5 MPa; Supplementary Table S1). Production water from the reservoir used in this study contained 6.4 mM of acetate, 39.6 mM of bicarbonate and indigenous microbes. We inoculated 8 ml of crude oil and 800 ml of the production water as a source of indigenous microbes and nutrients into the apparatuses, pressurized with either N2 or N2/CO2 (90:10) gas to a final pressure of 5 MPa, and incubated at 55 °C. The microcosms incubated under an N2 atmosphere were designated ‘control microcosms’, and those incubated under an N2/CO2 atmosphere were designated as ‘CO2-injected microcosms’. Control and CO2-injected microcosm experiments were prepared in triplicate. Two of the three microcosms were amended with either a trace amount of 13C-bicarbonate (final 400 μM) or [2-13C]-acetate (final 87.5 μM) to track the methanogenic pathway.

After 1 week of incubation, CO2 partial pressures and bicarbonate concentrations in the control microcosms stabilized at ~0.004 MPa (=0.04 atm) and 47.7 mM and in the CO2-injected microcosms at 0.2 MPa (=2.0 atm) and 86.4 mM; decrease in CO2 was due to dissolution into the production water. The pH levels in both microcosms were stable at ~8.2 and 7.3, respectively, during incubation. The conditions in the CO2-injected microcosms were likely to reflect the bicarbonate concentration and pH in previously reported CCS sites (Supplementary Table S1). In all microcosms, methane productions increased after at least 50 days of incubation, whereas methane productions in CO2-injected microcosms completed ~30 days earlier than the control microcosms (Fig. 1a). Acetate present in the original production water decreased concomitantly with increased methane production. Methane production and acetate degradation in microcosms labeled with stable isotope tracers were almost similar to unlabeled microcosms. Methane production was nearly equivalent to the acetate consumed. This stoichiometric methanogenic reaction suggests that almost all the methane formed was derived from acetate. Interestingly, the methane production rate in CO2-injected microcosms (0.36±0.04 μmol d−1 ml−1 water at the growing phase) was twice as high as in control microcosms (0.17±0.02 μmol d−1 ml water at the growing phase), indicating that CO2 injection stimulated methanogenesis.

Figure 1: Methanogenesis in control and CO2-injected microcosms.
figure 1

Methanogenesis shown in (a) three individual control microcosms and (b) three individual CO2-injected microcosms under high-temperature and high-pressure conditions (55 °C and 5 MPa). Methane (circles) and acetate (squares) concentrations over time in the unlabeled (closed symbols), [2-13C]-acetate (gray symbols) and 13C-bicarbonate (open symbols)-labeled microcosms. All six microcosms were constructed using high-temperature oil reservoir samples.

CO2-driven change in methanogenic pathway in microcosms

To identify the methanogenic pathways in the control and CO2-injected microcosms, the carbon isotopic compositions (δ13C) of methane and dissolved inorganic carbon (DIC) were measured during incubation. In the control microcosms labeled with 13C-bicarbonate, elevated δ13C methane values (>450‰) were detected (Fig. 2a), showing that the amended 13C-bicarbonate was converted to 13CH4. In the [2-13C]-acetate-labeled control microcosm, δ13CDIC values gradually increased and methane also became enriched in 13C following DIC (Supplementary Table S2), indicating that the 13C-methyl group of acetate was oxidized to CO2 and incorporated into a large CO2 pool, and a small fraction of 13CO2 was converted to 13CH4 (acetate oxidation: CH3COO+H++2H2O→4H2+2CO2; methanogenesis from H2: 4H2+CO2→CH4+2H2O; over all, CH3COO+H+→CH4+CO2). In contrast, the [2-13C]-acetate added to a CO2-injected microcosm resulted in elevated δ13C methane values (>850‰) (Fig. 2b). This indicates that the 13C-methyl group of acetate was directly converted to 13CH4 (13CH3COO+H+13CH4+CO2). These observations demonstrated that the dominant methanogenic pathway in the control microcosms was syntrophic acetate oxidation coupled with hydrogenotrophic methanogenesis, while acetoclastic methanogenesis was the dominant methanogenic pathway in the CO2-injected microcosms. The values of carbon isotopic fractionation between CH4 and DIC in the unlabeled control (εC=−83.3±1.3‰) and CO2-injected (εC=−10.9±1.0‰) microcosms during incubation (Supplementary Table S2) were also consistent with the hydrogenotrophic and acetoclastic methanogenesis, respectively16. The reproducibility of the impact of CO2 on the methanogenic pathway and community resiliency were confirmed by an additional stable isotope tracer experiment, which showed that the dominant methanogenic community and pathway switched from one to the other when the headspace gas of the control microcosm was replaced with N2/CO2 (90:10) and the CO2-injected microcosm was replaced with 100% N2 (Supplementary Note 1; Supplementary Figs. S2 and S3, and Supplementary Tables S3, S4 and S5).

Figure 2: Carbon isotopic ratios of methane in control and CO2-injected microcosms after incubation.
figure 2

Carbon isotopic ratios of methane in the (a) control microcosms and (b) CO2-injected microcosms after incubation of 103 days. Carbon isotopic ratios of methane and dissolved inorganic carbon (DIC) in each microcosm during incubation are shown in Supplementary Table S2.

CO2-induced alteration of methanogenic microbial community

Quantitative PCR (qPCR) and 16S rRNA gene clone library analysis showed differences in microbial community structure in the original production water compared with the incubated microcosms. The qPCR assays (Fig. 3a) of the archaeal 16S rRNA genes showed that (1) the original production water contained mostly the order Methanobacteriales and a smaller population of Methanosarcinales; (2) the control microcosm was similarly dominated by hydrogenotrophic methanogens affiliated with the order Methanobacteriales; whereas (3) the CO2-injected microcosms differed and were dominated by the order Methanosarcinales, many of which are acetoclastic methanogens. Clone library analyses supported the qPCR results. In the original production water and control microcosm, the genus Methanothermobacter dominated the archaeal community (Fig. 3b and Supplementary Table S4). The control microcosm bacterial community was dominated by a limited number of phylotypes that includes the genus Thermacetogenium17, a known syntrophic acetate-oxidizing bacterium (99.8% sequence similarity) (Supplementary Table S5). In contrast, all archaeal clones in the CO2-injected microcosms were affiliated with the acetoclastic species Methanosaeta thermophila strain PT (99% sequence similarity) (Fig. 3b and Supplementary Table S4). In addition, in the CO2-injected microcosms, heterotrophic fermentative bacteria such as Coprothermobacter proteolyticus dominated in the bacterial community and the known syntrophic acetate-oxidizing bacteria disappeared (Supplementary Table S5). These results clearly demonstrated that the main constituents of the control microcosm methanogenic microbial community were hydrogenotrophic methanogens and syntrophic acetate-oxidizers; CO2 injection changed the community and led to the dominance of acetoclastic methanogens. These findings are consistent with the change in methanogenic pathways assessed by the stable isotope tracer experiment.

Figure 3: Archaeal methanogenic community compositions.
figure 3

Archaeal methanogenic community compositions in the original production water, control (day 116) and CO2-injected microcosms (day 103). (a) The 16S rRNA gene copy concentrations of the domain Archaea (Arc), the orders Methanobacteriales (MBT), Methanomicrobiales (MMB), and Methanosarcinales (MSL) quantified by qPCR assays. ND, not detected. (b) Relative abundance of archaeal clones. MBT1: Methanothermobacter thermautotrophicus KZ3-1 (DQ657903) (>99% similarity), MBT2: Methanothermobacter wolfeii KZ24a (DQ657904) (>95% similarity), MSL1: Methanosaeta sp. (AJ133791) (99% similarity), MSL2: Methanolobus psychrophilus R15 (EF202842) (98% similarity), MSL3: Methanosaeta thermophila PT (AB071701) (99% similarity), MMB1: Methanoculleus receptaculi ZC-3 (DQ787475) (99% similarity), Other: Thermococcus litoralis DSM5474 (AY099180) (99% similarity). Details of archaeal clone library analysis are shown in Supplementary Table S4.

Thermodynamic constraints on methanogenic reactions

To elucidate the underpinning mechanisms of the shift in methanogenic pathways in conjunction with microbial community transition by injecting CO2, we conducted thermodynamic calculations for the following reactions under various partial pressures of CO2: acetate oxidation, hydrogenotrophic methanogenesis and acetoclastic methanogenesis. The thermodynamics showed that hydrogenotrophic methanogenesis becomes energetically more favorable with increasing CO2 partial pressure, whereas acetate oxidation and acetoclastic methanogenesis become less favorable (Fig. 4a). However, acetoclastic methanogenesis is less sensitive to high CO2 partial pressure than acetate oxidation because acetate oxidation produces two moles of CO2 per mole of acetate, whereas acetoclastic methanogenesis produces one mole of CO2. The ΔG values calculated for the three reactions in the control and CO2-injected microcosms showed that acetate oxidation in the CO2-injected microcosm would be endergonic (Fig. 4a and Table 1), indicating that syntrophic acetate oxidation coupled with hydrogenotrophic methanogenesis could not proceed under high CO2 partial pressures but acetoclastic methanogenesis could proceed. A cultivation experiment verified the effect of CO2 partial pressure on microbes; a pure syntrophic defined coculture17 of Thermacetogenium phaeum strain PB and Methanothermobacter thermoautotrophicus strain TM was also inhibited by increasing CO2 partial pressure (Supplementary Note 2 and Supplementary Fig. S4).

Figure 4: CO2 partial pressure effect on change in Gibbs free energy for acetate oxidation and methanogeneses.
figure 4

Effect of CO2 partial pressure on the change in Gibbs free energy for acetate oxidation, hydrogenotrophic methanogenesis and acetoclastic methanogenesis. (a) The setup conditions are at 116 days of incubation in the control microcosm: acetate at 0.3 mM, methane at 0.51 atm, hydrogen at 1.5 × 10−4 atm, pH=8.1, at 55 °C and 50 atm. (b) The setup conditions are at 77 days of incubation in the CO2-injected microcosm: acetate at 0.4 mM, methane at 0.66 atm, hydrogen at 1.2 × 10−4 atm, pH=7.3, at 55 °C and 50 atm. Dashed lines indicate actual CO2 partial pressures (0.04 and 2.0 atm) in the control and CO2-injected microcosms, respectively.

Table 1 Thermodynamic parameters and Gibbs free energies of three reactions in the control and CO2-injected microcosms.

Discussion

This study illustrates the possibility that an increase in CO2 partial pressure will change the microbial ecosystem in a deep subsurface high-temperature oil reservoir. Our results suggest that CO2 injection into deep subsurface oil reservoirs can alter the in situ methanogenic pathway and lead to the dominance of acetoclastic methanogenesis. Acetate is often the most abundant organic acid in oil reservoirs18,19 and an important intermediate in methanogenesis from crude oil20. Given that acetoclastic methanogenesis generally leads to a faster methanogenic reaction than syntrophic acetate oxidation coupled to hydrogenotrophic methanogenesis (Supplementary Note 2 and Supplementary Fig. S4), the shift to this pathway may result in an acceleration of methane production in oil reservoirs as observed in this study (Fig. 1a). Furthermore, the thermodynamics, in theory, suggests that the shift to acetoclastic methanogenesis can ease the thermodynamic constraint of crude oil biodegradation (Supplementary Fig. S5). Perhaps, the largest window of opportunity for crude oil biodegradation via acetoclastic methanogenesis alone21 may lend further support to the hypothesis that CCS may increase energy recovery in the form of methane through crude oil biodegradation. Our results present a possibility of CCS for enhanced microbial energy production in deep subsurface environments that can mitigate global warming and energy depletion at the same time. To date, very little is known about the influence of CO2 sequestration on the subsurface microbial communities and their functions, despite their important contribution to global biogeochemical processes22,23. This study would intrigue not only geochemists but also microbiologists for further investigation of CCS in connection with utilization of microbial activities in deep subsurface.

Methods

Study site and sampling

Yabase oil field is one of the oldest and largest onshore oil fields in Japan, located in Akita Prefecture (39° 42′N, 140° 5′E). Currently, the oil field is almost depleted and characterized by high overall water cut (~90% basic sediment and water). The main reservoir rocks are tuffaceous sandstone of Miocene–Pliocene age. The depth of the oil horizon ranges from 1000 to 1300, m, with in situ temperature and pressure estimated to be 53–65 °C and 5 MPa, respectively.

In Yabase oil field, there are some production wells in which crude oil along with production water has been produced by pumping. In this study, reservoir samples were collected from one of the wells where water injection to enhance oil recovery has never been applied. The samples were taken from production flow at the wellhead by discharging the fluid mixture through a metal tube into gas-tight glass bottles flushed in advance with nitrogen gas in September 2009. Immediately before each bottle was sealed, the headspace was flushed and pressurized with nitrogen gas to minimize air contamination. The samples were maintained at 50°C for 3 days until use.

Geochemical characteristics of the production water and gas sample from this reservoir have been previously described in the study by Mayumi et al.15 Briefly, the water had low salinity (Cl; 4690, mg l−1) and −239 mV of reduction potential, and concentrations of nitrate and sulfate as electron acceptors were 14.5 μM and below 0.5 mM, respectively. Gas compositions of the sample collected from the production flow were H2: 0.1%, CO2: 2.4%, CH4: 77.2%, C2H6: 10.8%, C3H8: 5.2%, i-C4H10: 0.8% and n-C4H10: 1.5%.

Microcosms

Six microcosms were prepared with 800 ml of production water and 8 ml of crude oil in 1 l sterile stainless-steel cylinder bottles (304L-HDF4-1000; Swagelok, Ohio, USA) and stable isotope tracers were added as described in Supplementary Fig. S1. Control microcosms were pressurized with nitrogen gas, and CO2-injected microcosms were pressurized with N2/CO2 (90:10; δCO2=−34.8‰) at 5 MPa. Before pressurization of CO2-injected microcosms, sodium bicarbonate (δNaHCO3=−4.2‰) was added to a final concentration of 74 mM to mimic in situ conditions where formation water is highly buffered due to the presence of minerals such as calcite (CaCO3) and dolomite (CaMg(CO3)2) in oil reservoirs9,12. The microcosms were incubated at 55 °C, and the mixing ratios of CH4 and CO2 in the headspace gases and organic acids in production water were periodically measured with a gas chromatograph and an ion chromatograph15. H2 concentrations in the headspace gases were measured with an EAGanalyzer gas chromatograph equipped with a semiconductor detector (Sensortec Co., Ltd., Shiga, Japan).

Carbon isotope analysis

Headspace gas and the incubated water were collected into gas-tight glass cylinder bottles and vials, respectively, from all microcosms periodically. Carbon isotopic compositions of CH4 in gas-tight glass cylinder bottles were determined with a Finnigan gas chromatograph combustion isotope ratio mass spectrometer (GC-C-IRMS) consisting of a Hewlett Packard 5890 GC, a Finnigan MAT 252 IRMS and a ThermoQuest combustion interface (Thermo Finnigan Inc., Texas, USA). Carbon isotopic composition of DIC in the incubated water was measured after the addition of 1 M H2SO4 to liberate total CO2. All measurements were conducted in triplicate, and the s.e. values were less than 1‰. The isotopic values were expressed in δ notation relative to the Vienna Pee Dee Belemnite (VPDB) standard. The values of isotopic fractionation (εC) between DIC and CH4 were determined by εC=(δCH4−δDIC)/(1+δDIC/103).

Molecular biological analyses

Molecular analyses were carried out for the original production water and incubated water from [2-13C]-acetate labeled microcosms (at day 116 and 103 for the control and the CO2-injected microcosms, respectively). Total DNA was extracted from a 0.22-μm-pore-size polycarbonate membrane filter (Millipore, MA, USA), concentrating 250 ml of the original production water, 15 ml of the control microcosm water and 40 ml of the CO2-injected microcosm water, using a FastDNA Spin kit (MP Biomedicals, CA, USA) according to the manufacturer’s protocol. Bacterial and archaeal 16S rRNA genes were amplified with primer sets Eub8F/Univ1490R24 and Ar109F25/Univ1490R, respectively. PCR products were cloned using the pT7 Blue T-vector kit (Novagen, CA, USA). Clones were randomly selected (in archaeal libraries, original production water; 32 clones, control microcosm; 39 clones, CO2-injected microcosm; 35 clones, in bacterial libraries, original production water; 49 clones, control microcosm; 92 clones, CO2-injected microcosm; 84 clones), and the inserted 16S rRNA gene was directly amplified with T7 promoter primer and U-19mer primer (Novagen). Sequence analysis of 16S rRNA genes was carried out with a BigDye Version 3.1 reaction on an ABI3730xl DNA Analyzer (Applied Biosystems, CA, USA). Chimeric sequences were detected with the Bellerophon server26 and removed from the sequence data sets. To group the OTUs and draw rarefaction curves, sequences were analyzed using FastGroupII program27 with 97% sequence similarity. Representative sequences from each OTU were compared with those in public databases using the Seqmatch program from the Ribosomal Database Project (Release 10, Update 12) to identify the nearest neighbors.

The specific primer and probe set for archaea, hydrogenotrophic methanogens (the orders Methanobacteriales and Methanomicrobiales), and acetoclastic methanogens (the order Methanosarcinales) used in quantitative real-time PCR to determine the 16S rRNA gene copy numbers were previously designed in the study by Yu et al.28 All real-time PCR assays were performed with iCycler iQ using the iQ Supermix reaction kit (Bio-Rad, California, USA), as previously described in the study by Mayumi et al.15 Standard curves for each assay were constructed using nearly full-length 16S rRNA gene fragments amplified from Methanothermobacter thermautotrophicus strain delta H (DSM 1053) for the Archaea and the Methanobacteriales assays, Methanoculleus bourgensis strain MS2 (DSM3045) for the Methanomicrobiales assay, Methanosaeta thermophila strain PT (DSM 6194) for the Methanosarcinales assay.

Thermodynamics calculations

Gibbs free energy calculations were made according to the study by Dolfing et al.21 Temperature corrections were made with the Gibbs–Helmholz equation according to ΔGoTact=ΔGoTref.(Tact/Tref)+ΔHoTref·(TrefTact)/Tref with T in K; Tact=328.15 K, Tref=298.15 K. The effect of pressure on ΔG (in kJ per reaction) was approximated as ΔGoTref,Pact=ΔGoTref,Pref+ΔVo·(Pact−Pref)/10000 where ΔGoTref,Pact is the Gibbs free energy of reaction at the reference temperature (298.15 K) and in situ pressure (5 MPa; 50 atm), ΔGoTref,Pref is the standard Gibbs free energy of reaction at the reference temperature and pressure (0.1 MPa; 1 atm), Pact is the in situ pressure in atm, Pref is the reference pressure in atm and ΔVo is the partial molar volume change of the reaction at the reference temperature and pressure in cm3 mol−1, as outlined in the study by Wang et al.29. The calculations were made for acetate (pKa=4.75) at the appropriate pH values30. For the reaction CO2+4H2→CH4+2H2O, ΔVo=−62.3; for CH3COO+H++2H2O→4H2+2CO2, ΔVo=92.25; for CH3COO+H+→CH4+CO2, ΔVo=29.95. Neglecting activity coefficients yields an error in calculated Gibbs free energy values of at most 2 kJ mol−1 for all reported values at an ionic strength of 0.1 M (Debye–Huckel calculations).

Additional information

Accession codes: All 16S rRNA gene sequences obtained in this study have been deposited at DDBJ under accession numbers AB668482AB668513 and AB710350AB710376.

How to cite this article: Mayumi, D. et al. Carbon dioxide concentration dictates alternative methanogenic pathways in oil reservoirs. Nat. Commun. 4:1998 doi: 10.1038/ncomms2998 (2013).