On sorption and swelling of CO2 in clays
- 2k Downloads
The geological storage of carbon dioxide (CO2) is a well-studied technology, and a number of demonstration projects around the world have proven its feasibility and challenges. Storage conformance and seal integrity are among the most important aspects, as they determine risk of leakage as well as limits for storage capacity and injectivity. Furthermore, providing evidence for safe storage is critical for improving public acceptance. Most caprocks are composed of clays as dominant mineral type which can typically be illite, kaolinite, chlorite or smectite. A number of recent studies addressed the interaction between CO2 and these different clays and it was shown that clay minerals adsorb considerable quantities of CO2. For smectite this uptake can lead to volumetric expansion followed by the generation of swelling pressures. On the one hand CO2 adsorption traps CO2, on the other hand swelling pressures can potentially change local stress regimes and in unfavourable situations shear-type failure is assumed to occur. For storage in a reservoir having high clay contents the CO2 uptake can add to storage capacity which is widely underestimated so far. Smectite-rich seals in direct contact with a dry CO2 plume at the interface to the reservoir might dehydrate leading to dehydration cracks. Such dehydration cracks can provide pathways for CO2 ingress and further accelerate dewatering and penetration of the seal by supercritical CO2. At the same time, swelling may also lead to the closure of fractures or the reduction of fracture apertures, thereby improving seal integrity. The goal of this communication is to theoretically evaluate and discuss these scenarios in greater detail in terms of phenomenological mechanisms, but also in terms of potential risks or benefits for carbon storage.
KeywordsCCS CO2 storage Clay swelling Carbon dioxide Leakage Containment Smectite
For the characterization of geological CO2 storage reservoirs, a number of critical parameters need to be assessed. From existing knowledge and experience, especially collected in the oil and gas industry, the storage capacity and injection rate are generally well understood for specific reservoirs. Critical parameters are reservoir size and reservoir heterogeneity, i.e. porosity and (relative) permeability, fluid saturation, the reservoir stress field, in particular the minimum horizontal stress, as well as pressure and temperature conditions.
In addition, the identification and risk assessment of potential leakage pathways, reservoir depletion rate in case of leakage and reservoir pressure at which leakage is initiated or inhibited, are unique to a CCS project and thus need to be considered. Pressure is a key parameter in any leakage scenario, and will decrease in typical fluid extraction processes but increase in storage applications. The CO2 injected into a reservoir may, sometimes significantly, increase the average reservoir pressure. Initially, pressure builds up only locally, i.e. in the vicinity of the injection well. Imperfections in cementation of injection, monitoring or abandoned wells, can result in the formation of micro-annuli between cement and caprock or cement and casing, potentially acting as pathways for gas leakage. With continuing injection, the pressure pulse will eventually be transmitted to the far field. The resulting rise of the reservoir pore pressure reduces the effective stress on existing fractures and faults, potentially causing their (re)activation.
The main trapping mechanisms in CO2 storage are structural and residual trapping. In structural trapping, a continuous, connected gas column will form underneath a sealing formation. Buoyancy results in fluid pressure acting on the reservoir-caprock interface, which must be lower than the capillary entry pressure of the seal to prevent capillary leakage. Hence, this maximum pressure or gas column height is a key parameter in the assessment of a storage scenario. In residual trapping, gas resides in disconnected bubbles in pores and is therefore not contributing to a buoyancy pressure.
The factors controlling leakage from a gas storage reservoir have been addressed earlier for a range of applications. Yet, the specific properties of CO2 add to the complexity of the assessment of leakage scenarios. For instance, CO2 dissolves in brine or water, forms a weak acid and possibly reacts with surrounding rock surfaces. While many of these reactions are probably rather insignificant within the time scales of interest (<10,000 years), the dissolution or precipitation of carbonates and sulfates might occur within relevant time scales. In addition, CO2 exhibits specific wetting behaviour, and incomplete water-wetting conditions have been reported. Moreover, wetting properties may be altered by water–rock-interactions (Iglauer et al. 2015), potentially affecting two-phase flow in the reservoir and capillary sealing of barriers.
CO2 also interacts with the nanopores of shales, which largely consist of clay minerals, by diffusion through the aqueous phase or drainage of water from the pore space and subsequent adsorption on the high surface area clays. In the last 5–10 years, a series of studies (Busch et al. 2008; Loring et al. 2011; Rother et al. 2013a; Schaef et al. 2012; Weniger et al. 2010, amongst others) reported on this interaction, with a major focus on swelling clays, such as montmorillonite (MMT). It was found that for CO2 storage containment, non-negligible physical effects result from reacting clays with CO2 at hydrostatic pressure, temperature, and stress conditions representative of geological reservoirs.
The aim of this contribution is to provide a summary of these studies, and to put them into perspective with regards to the geological storage of CO2. We present a brief literature review, followed by detailed discussions of several storage scenarios in which CO2-clay interactions may play a role and affect sealing efficiency of caprocks and wells. The potential of CO2-clay interactions for CO2 trapping is discussed. Temperature and pressure changes in the reservoir affect local and reservoir stresses. These are standard reservoir management issues relevant in many different improved or enhanced production operations and need to be addressed in the geomechanical evaluation of any Carbon Capture and Storage (CCS) project.
1.1 Clay minerals in geological reservoirs
Before addressing the interactions between CO2, water, and clay minerals, with a specific focus on smectite, it is helpful to review the stability of clays in the sub-surface at the specific pressures, temperature and chemical conditions relevant for geological reservoirs. A general overview on clay classification and physico-chemical properties is provided in Bergaya et al. (2006), amongst others. Smectite is a low-charge, expandable clay. Its basic structure consists of a 10 Å layer composed of an octahedral sheet sandwiched between 2 tetrahedral sheets (TOT structure), plus the pore space (interlayer space) between such TOT structures. This interlayer space is expandable, depending on hydration state, i.e. the amount of water layers confined in this space. During sediment burial, smectite commonly convert into illite by passing through a transition phase characterized by smectite-illite mixed layers. Different models have been discussed for this reaction, and an overview of this work is provided in e.g. Altaner and Ylagan (1997). The smectite-to-illite transition depends on a range of factors like temperature, geologic time, porosity and permeability, connectivity between formations, formation water chemistry, and water/rock ratio. All these factors are oftentimes constantly changing with basin history. The entire process is complex but has important implications for many aspects like geo-pressurization, the growth of fault and fracture zones, or hydrocarbon migration. More details are given in Altaner and Ylagan (1997), Brigatti et al. (2013), Ferrage et al. (2011) and Lanson et al. (2009). Clay minerals (e.g. illite, smectite, kaolinite, chlorite) typically occur in every sedimentary basin, no matter what the reservoir conditions are. Expansion of clays was, however, mainly observed for smectite (exceptions apply), and we therefore briefly discuss the stability of smectite in relation to the above mentioned smectite-to illite-transformation. In geological reservoirs the controlling factor for the transition from pure smectite to illite, through an illite/smectite (I/S) mixed-layer transition stage, can be temperature. Several studies investigated the relative smectite content in I/S mixed-layers, and found that for a sedimentary basin with a geothermal gradient of 25–35 °C/km, smectite content decreases to 20 % or even less, assuming that at temperatures of ~120–175 °C the transformation is complete. This corresponds to a burial depth of at least 4–6 km (e.g. Hower et al. 1976; Lanson et al. 2009; Velde and Vasseur 1992) and gives constraints on the occurrence of swelling clays in sedimentary basins, and hence within tight argillaceous rocks overlying CO2 storage reservoirs. Especially deeply buried reservoirs, that underwent inversion resulting in present day depths of 1–3 km, will have had most or even all of their initial smectite content transformed to illite. Such depths are considered to be good candidates for CO2 storage due to high CO2 densities at moderate depths, keeping drilling costs low. A summary of clay mineralogy and smectite contents of the different CO2 storage operations around the world is given by Espinoza and Santamarina (2012). Clay contents, when mudrocks represent the primary caprock, are usually 50 % and higher. Smectite contents however are usually small with values of up to 9 % (Sleipner, Norway), or 1–3 % (Otway, Australia). High I/S mixed layers contents have been reported for the SACROC (up to 62 %) and Frio (45 %) projects, both in the USA (Espinoza and Santamarina 2012). Smectite contents within the mixed-layers were not determined. Irrespective of these observations, young subsiding basins, like many reservoirs in the North Sea or the Gulf of Mexico, can show significant amounts of smectite or I/S mixed layers, and are hence subject to interactions with CO2 as will be discussed in the following.
1.2 CO2 sorption on clay minerals
An alternative explanation is that the excess sorption capacity of clays is generally not related to the large surface area of the interlayer space, but only to the mesopores between clay particles. While this may be true for nitrogen, accessibility of the interlayers of hydrated smectites to at least some fluids, including water and carbon dioxide, has been shown.
CO2 sorption isotherms of clays and shales (but also other materials) consistently show that the excess sorption decreases after passing through a maximum, and sometimes even becomes negative at high fluid densities above the critical point. This maximum coincides with the steep increase of the CO2 density curve around the critical density of CO2 (7.3 MPa). It is therefore lower than the pressure in reservoirs typically considered for CO2 storage. Possible reasons for this have been discussed elsewhere; see for instance Rother et al. (2013a, b). Excess sorption describes the advantage of adsorbing gas (in this case CO2) over storing it in the bulk phase. The difference between bulk density and pore density is the storage density in the accessible pore spaces. If the excess sorption is positive, the sorption phase is denser than the coexisting bulk fluid phase, and when excess sorption is negative it is more efficient to store CO2 in the bulk pore volume. Consequently, when the excess sorption is zero, both densities are equal. Experimental issues can further complicate the pore storage efficiency assessment, as discussed in Siemons and Busch (2007) or Busch and Gensterblum (2011). Rother et al. (2013a) explicitly addressed the density of the sorbed phase. They report a combined excess sorption and neutron scattering study on Texas montmorillonite (STx-1) to investigate sorption amounts, interlayer swelling and sorbed phase density at temperatures of 35 and 50 °C and pressures up to 15 MPa. They found that at pressures below the critical CO2 pressure (P c = 7.3 MPa) the sorbed CO2 density is higher than the gas density. At about 8–10 MPa equal bulk and sorbed phase densities of ~300–400 kg/m3 were observed. With further pressure increases, the bulk density becomes increasingly larger than the sorbed phase density, hence resulting in negative excess sorption. Similar observations have been made in earlier studies by the same authors on silica gels with defined pore geometries (Rother et al. 2013b). These findings partly confirm those by Busch et al. (2008) where a decreasing excess sorption trend was observed for pressures exceeding ~8 MPa, but negative excess sorption (where the sorbed phase density is below bulk density) was not observed. Negative and near negative excess sorption of CO2 on clays and shales was however reported by Busch et al. (2012). The origins of high-density interfacial fluid depletion are molecular, i.e., weakly attractive fluid solid interactions, causing fluid–solid interactions to be energetically preferred over fluid–solid interactions at low density, but fluid–fluid interactions are preferred at high density.
1.3 Sorptive swelling of clays with CO2 exposure
Besides water content, swelling strain was also shown to depend on the interlayer cation. Giesting et al. (2012a, b) have demonstrated that the swelling strain from Na and K-exchanged samples is similar with a return to initial values at discrete water layers. This seems to be different for their Ca-exchanged equivalents that show higher swelling strains and only seem to return to initial values at a 2 W hydration state.
To transfer this information to subsurface conditions, it is essential to know the hydration state at different depths. As mentioned earlier, Bird (1984) calculated hydration states with burial depth from thermodynamic considerations, showing rough depth ranges where Ca or Na-exchanged smectites are rather in the 1 or 2 W hydration state. Unfortunately, Bird’s calculations are on a basin scale, i.e., precise clay hydration states were not defined, and it remains unclear whether discrete or intermediate states are to be expected in a geologic reservoir. Different speculative models were developed for the mechanism of the observed swelling: Giesting et al. (2012b) discussed the configuration of CO2 in the clay interlayer, suggesting the formation of a permanent carbonate species. This assumption was based on the observation that K and Ca-Wyoming smectite did not return to their respective original hydration states after a pressure cycle, i.e., a hysteresis remained. This effect became more pronounced with longer charging times. Permanent trapping of CO2 in the interlayer of Na-exchanged smectite (Na-SWy-2), potentially as carbonates, was also indicated in the work by Hur et al. (2013) and Romanov (2013) from Fourier-transform infrared spectroscopy (FTIR) and X-ray diffraction data. However, CO2 trapping in clays was not observed in a number of other studies: Krukowski et al. (2015) using FTIR on Na-exchanged (Na-STx-1), and Schaef et al. (2012) on Ca-exchanged (Ca-STx-1) Texas-montmorillonite using thermo gravimetric analysis, as well as Loring et al. (2012), also on the same Ca-exchanged samples using a variety of methods, such as magic angle spinning nuclear magnetic resonance spectroscopy and attenuated total reflection infrared spectroscopy. These measurements were performed at 50 °C and pressures up to 18 MPa.
Loring et al. (2014), using X-ray diffraction and IR spectroscopy at 50 °C and 9 MPa on Na-SWy-2, studied relative CO2 uptake as a function of clay water content, and demonstrated a steep increase in CO2 interlayer contents at low water saturations. This steep increase seems to correspond to a step in hydration from 0 to 1 W. With an increase in water content, this CO2 concentration decreases and almost disappears when the clay interlayer distance moves to 2 W. This clearly demonstrates that the interlayer CO2 uptake capacity is strongly related to the water content. It also confirms the observation that for Na-exchanged smectite (Giesting et al. 2012a) interlayer swelling occurs between 0 and 1 W only.
In a very recent study Schaef et al. (2015) demonstrated, using XRD, IR and quartz crystal microbalance (QCM) on variably hydrated Ca-SWy2, that the swelling depends on the relative amounts of H2O and CO2 in the interlayer. The interlayer space of dry smectite increases sharply upon introduction of some water, and decreases again upon further hydration of the clay. These findings qualitatively confirm observations discussed above and demonstrate that clay swelling is feasible, when the interlayer distance is between 0 and 2 W. These states correspond to subsurface or reservoir conditions (e.g. Bird 1984), but shrinkage occurs for hydration states >2 W, i.e., at near-surface to surface conditions, likely due to the removal of water by dry scCO2.
1.4 Swelling stresses induced by CO2 sorption to clays
Values for effective stresses depend on the pore pressure Pp, the swelling pressure Sp, and the lithostatic pressure. Wentinck and Busch (2014) have shown that for specific CO2 storage scenarios, where the seal contains high amounts of swelling clays, the risk of shear failure is given for time scales in the order of 100s or 1000s of years. For shear failure to occur, a number of factors, like swelling clay content, effective stress conditions and related initial smectite interlayer spacing, diffusion coefficient controlling velocity of CO2 penetrating the caprock, stress relaxation into the shale formation etc. have to be favourable for the development of significant swelling stresses. It remains difficult to predict if, in case of shear failure, a permeable pathway would develop.
This clearly is an important area for future research and critical parameters should be constrained, especially when it comes to fault behaviour and build-up and relaxation of stress following clay swelling.
1.5 Molecular dynamics studies
A limited number of molecular dynamics (MD) and grand canonical Monte–Carlo (MC) simulation studies aimed at the explanation of the experimental results presented above. In a combined MD and MC study, Botan et al. (2010) found for pressure and temperature conditions representative for CO2 storage reservoirs, that at least one CO2 molecule per clay unit cell is capable of entering hydrated clay interlayers, hence qualitatively confirming the above experimental results. They did, however, not confirm the strain response upon CO2 intercalation (neither swelling nor shrinkage). More recent MD studies using density functional theory demonstrate, that CO2 is able to intercalate the clay mineral interlayer resulting in positive strain (Cygan et al. 2012; Myshakin et al. 2013, 2014; Spiering et al. 2014), and that the strain strength depends on the initial hydration state. These simulations were typically carried out at p, T representing supercritical CO2 and therefore reservoir conditions (P c > 7.39 MPa, T c > 31 °C). It was shown that swelling increases with an increase in the number of water molecules, and with an increase in the number of CO2 molecules per unit cell. It was also shown that different initial water/CO2 molecular ratios lead to stable final states (0, 1 W, etc.). While this in general confirms the experimental observations, some discrepancies remain. In their systematic studies, Giesting et al. (2012a, b) showed that for the different cation-exchanged Wyoming smectite (SWy-2), swelling is most dominant just above the stable hydration states of 0, 1 W etc. This shows first of all that some interlayer water is needed to swell the sample (no swelling at 0 W), and that smectite at discrete hydration states (e.g. 1 W) do not exhibit major swelling. Between these defined states, the swelling strain increases initially, followed by a decrease with increasing hydration (cf. Figure 2). Spiering et al. (2014) argued that CO2 acts as a “catalyst”, promoting increased hydration of the interlayers. For the CO2 case, this indicates that the swelling strain is larger compared to the non-CO2 case at the same hydration state. Wentinck and Busch (2014) confirm this interpretation, and explain the swelling mechanism by the contribution of the entropy of mixing between H2O and CO2 to the chemical potential of the interlayer space. This contribution, albeit only in the 100 s of Joule/mole-range, is large at low water content in the interlayer. However, it is critical as it results in cation hydration at lower relative humidity, i.e., clay hydration, compared to the non-CO2 case. This hypothesis is in agreement with Loring et al. (2014) who showed an initial steep increase in CO2 interlayer concentrations with increasing water contents in the range between 0 and 1 W, followed by a decrease towards higher hydration states.
2 Potential CO2 leakage scenarios associated with clay swelling
2.1 CO2 sorption on clay mineral surfaces in the reservoir
Gas adsorption on mesoporous materials is a well-known phenomenon in general and it is understood that the physical sorption capacity is proportional to the accessible SSA. SSA of typical reservoir rocks can vary by orders of magnitude. SSA of sand (2–0.05 mm) and silt size particles (0.05–0.002 mm), calculated assuming perfect spheres, vary between 0.001–0.04 and 0.04–1.1 m2 g−1, respectively. In contrast, clays exhibit a much larger surface area. Commonly reported values from N2-BET are 5–15 m2 g−1 for kaolinite, 25–40 m2 g−1 for illite, 5–15 m2 g−1 for chlorite and 80–120 m2 g−1 for smectite (Meunier, 2005), while the total smectite surface areas including the interlayer space can reach values of up to several hundred m2g−1. As different analytical methods often give different SSA values, particularly for smectites, it seems that the area strongly depends on the method used and the respective measuring conditions using different gases, temperatures and pressures. Nitrogen physisorption, the most commonly used analysis for surface area determination, is most meaningful for supermicropores, mesopores and smaller macropores (Bertier et al. in press). Areas from CO2 physisorption at 273 K yield considerably higher surface area values for smectites (Thomas and Bohor 1968), i.e. CO2 provides information on ultramicropores but limited information on larger pores. Nevertheless, the SSA of clays are several orders of magnitude higher than those of quartz, feldspar and most other common reservoir rock minerals, and even a few percent of clay will increase the bulk rock surface area significantly.
Reservoir parameters and mineralogy of the hypothetical reservoir used in this calculation
Geothermal gradient (ϑ)
Hydrostatic gradient (γ)
Water saturation (Sw)
Langmuir parameters for clay sorption
Langmuir pressure (VL)
Langmuir pressure (PL)
Sorption and structural trapping are complementary processes, though they do take place in the same pore spaces. In Fig. 5, sorption capacity is calculated in absolute amounts. Note that, depending on its physical characteristics (density and volume), a sorbed layer will fill a certain volume of the pore space. Consequently, the pore space accessible to bulk CO2 is reduced. As long as the sorbed phase density has a higher value compared to the bulk density, any storage capacity will benefit from this trapping mechanism, since some of the injected CO2 will not contribute to pressure-buildup and increase the overall storage capacity. However, it was shown in several studies (e.g. Busch et al. 2012; Rother et al. 2013a) using manometric and gravimetric sorption devices, and neutron diffraction techniques, that the excess sorption capacity can become negative at pressures exceeding ~10 MPa. This means that the sorbed layer has an average density smaller than bulk CO2, and therefore the same amount of fluid in the sorbed layer will occupy a larger volume as compared to the bulk. A practical consequence is that injection of the same quantity of CO2 will result in higher pressures in a reservoir with negative excess sorption capacity.
2.2 CO2/clay mineral interaction at the reservoir-seal interface
In summary, CO2 sorption on clay minerals in shale formations will increase flux rates after CO2 breakthrough, while times scales for breakthrough are still far above the critical time scale of 10,000 years requested by most regulators. At the same time, depending on the details of the CO2 concentration gradients across the seal, significant amounts of CO2 will be temporarily immobilized, which contributes to storage safety and to a reduction in reservoir pressure.
2.3 Implications for shear failure
The likelihood for the reactivation of pre-existing faults predominantly depends on current stress states. Faults are critically stressed when the state of stress equals the Coulomb faulting criterion. This mainly depends on fault orientation with respect to the direction of maximum stress (Barton et al. 1995). Samuelson and Spiers (2012) provided experimental evidence that the fault friction properties are not affected significantly by CO2, i.e., only critically stressed faults are able to slip when storing CO2. It was also shown, that only mechanically active faults can become hydraulically active or permeable (Townend and Zoback 2000), and the risk of loss of containment exists only for permeable faults.
Wentinck and Busch (2014) performed a numerical study of the potential for shear type failure in a smectite rich caprock around a fault offset. The smectite content was considered to be 30 %, which is typical for many North Sea regional caprocks (Pearson 1990). In the study, transport was modeled by aqueous diffusion only, at a rate of Deff ~ 10−10 m2s−1, and clay sorption, swelling strain, and swelling stress data from literature sources summarized earlier in this paper were used. The reservoir conditions studied were 20 MPa and 353 K. It was shown that the swelling pressure affects the effective pressure such that the shear capacity (cf. Eq. 1) exceeds a value of 1, potentially resulting in shear-type failure. This process was modeled as a function of a diffusional front migrating through the caprock. Migration by diffusion was estimated to be on the order of mm or cm per year. When the diffusional front has migrated far enough into the caprock (on the order of meters—after 100’s to 1000’s of years), the shear capacity utilisation of the caprock can be exceeded. This is because clay swelling results in significant stress build-up, affecting shear stresses, and eventually leading to mechanical failure. In the case that this mechanical shear failure creates a permeable path, loss of containment would be the consequence.
Many relevant aspects for the estimation of shear-type failure due to CO2/clay interaction are still not well understood. This is the case for the exact swelling strain and swelling stress that will evolve and largely depends on the fluid composition, layer charge, and cation identity inside the smectite interlayer spaces (see discussion above). Although Bird (1984) has estimated the hydration state of different cation-exchanged smectites for different burial depths, calculations are rather rough, i.e. on the basin scale but not on the reservoir scale. Hence, clay fractions could be in stable or discrete hydration states, and pressure build-up could therefore be non-existent or limited. Furthermore, it remains largely unknown as to what extent relatively immature, smectite-rich mudrocks will relax such swelling pressure by inherent plasticity (or ductile creep), especially over time scales needed to establish a significant pressure profile. In addition, the caprock thickness needed to create high permeability cracks or to reactivate fault remains uncertain. Even in the case fault permeability would develop, limited information is available on potential flow rates.
In summary, there is a chance of clay swelling leading to significant swelling stresses, causing shear-type failure and potentially loss of containment along a permeable fault. However, many relevant parameters need to be quantified with a reasonable level of confidence on a case by case basis in order to determine the actual risk of loss of containment.
2.4 CO2 in natural or induced fractures
2.5 Leakage along the wellbore-seal interface
2.5.1 Formation dehydration around a wellbore
Wells might have either been used for production that were drilled within different decades in the past or might be drilled specifically for injecting or monitoring CO2, using latest drilling and completion technologies. For example, a former production well might have experienced higher temperatures from produced fluid as compared to the injected CO2. This has consequences on contraction and expansion of the cement sheath and the surrounding host rock, and a small annulus of micrometer diameter can form between cement and host rock (e.g. Bois et al. 2009; Gasda et al. 2004; Loizzo et al. 2011). Although such an annulus is unlikely to form in case of a competent cement job, it cannot be ruled out. If pressure gradients are formed inside the well, which is likely during the injection period, significant cooling of the well caused by the Joule–Thomson effect may occur, especially when injecting into depleted reservoirs with high pressure gradients. This wellbore annulus can develop apertures in the order of 10–300 μm (Bachu and Bennion 2009), especially when cold CO2 is injected. Thermal shrinkage of the cement and casing steel as well as surrounding shale can occur. The thermal coefficients are 9 × 10−6 K−1 for Portland cement (Barlet-Gouedard et al. 2009), and between ~10−4 K−1 to 10−5 K−1 for shale, depending on orientation, water saturation and shale type and characteristics (Mohajerani et al. 2012, 2014; Monfared et al. 2011; Wang et al. 1996). Especially quartz-rich formations are sensitive to thermal cracking, which is due to the large differences in the thermal expansion coefficient of quartz in comparison to other minerals such as clays. Another scenario is the shrinkage of montmorillonite in clay rich formations around the wellbore with hot fluid production. This is even more the case if it has taken place over long periods of time (e.g. Dusseault 2011) or an increased water uptake capacity in swelling clays when cold fluids (CO2) are injected. Shrinkage of swelling clays may lead to local and temporal pore pressure changes due to redistributions of the void space and mineral volume in a pore. This is the case when the pressure pulses are not transported to the far field. As a consequence of thermal shrinkage due to cooling some CO2 might leak along an annulus that has potentially formed and extends to the surface, or into overlying, shallower strata. In the case of dry CO2 leakage, shale desiccation is reasonable to assume, which might have implications on CO2 transport and sorption in the host rock. In this case, cracks could form, allowing CO2 to enter the formation laterally at increased flow rates, changing from pure aqueous diffusion to viscous flow. The rates of diffusive water flux from the formation towards the annulus against the water uptake capacity of CO2 leaking along the annulus to determine the dominant transport mechanism (see Fig. 4, case 5 for an illustration). We here consider a scenario where the CO2 density at 2000 m depth (20 MPa, 80 °C) is ~637 kg/m3 (Span and Wagner 1996). Under these p,T conditions, CO2 dissolves ~1 mol ‰ H2O (Spycher et al. 2003). Other important parameters are the diffusion coefficient (10−10 to 10−12 m2/s, e.g. Busch et al. 2008; Schlömer and Krooss 1997, 2004), concentration gradient for diffusing species, the contact area between annulus and shale formation (or the annulus length), and the annulus aperture itself (10–300 μm). The pressure gradient Δp is 20 MPa in the case of a direct connection to the surface, and assuming no overpressure from injected CO2 (which is likely). The gradient is smaller when CO2 is leaking into permeable reservoirs at shallower depth. In order to perform a rough mass balance calculation between water in-flow rates compared to uptake rates by the CO2, we use Fick’s Law to calculate diffusional fluxes of water to the annulus (Eq. 7). If we assume the effective diffusion coefficient D eff to be 1.0 × 10−10 m2 s−1, the concentration difference Δc to be 6.1 mmol.mol−1 (~100 mol H2O/m3 CO2, e.g. Spycher et al. 2003), and the distance Δx to be 0.1 m (for which it takes the CO2 about 3.2 years to diffuse to—so well within the timescale of a CO2 storage operation), we obtain flux rates of ~0.1 μmol m−2 s−1 from the formation to the wellbore annulus. Assuming an annulus diameter of 0.2 m and a caprock thickness of 100 m, we calculate diffusional flux rates of ~6 μmol H2O per second from the host rock to the wellbore.
These calculations suggest that shale desiccation in a well having an annulus between cement and rock is realistic, while the extent of which depends on a few parameters that need to be obtained in order to get quantitative results. Progressive desiccation might result in increased caprock accessibility for CO2, and therefore an increased potential for sorption and swelling, but also for shrinkage. If the shale contains large amounts of swelling clays, and dry CO2 is able to remove some or all of the interlayer water, desiccation cracks are likely to form, that will even further increase communication between shale and wellbore. This, however, is only likely in the case that the leaking CO2 has a continuous communication to either shallow aquifer or the surface. When CO2 in the annulus becomes stagnant, saturation with formation water can be expected to happen quickly, resulting in a limited shale desiccation.
2.5.2 CO2 sorption/swelling of clays around wellbore annulus
3 CO2 storage in (depleted) gas shale reservoirs
A number of articles discuss the possibility of storing CO2 in depleted gas shale reservoirs, or to use CO2 for enhancing shale gas production (e.g. Edwards et al. 2015; Khosrokhavar et al. 2014; Li and Elsworth 2014; Tao and Clarens 2013; Godec et al. 2013; Kang et al. 2010; Liu et al. 2013; Nuttall et al. 2009). These concepts are similar to CO2 enhanced coalbed methane recovery (CO2-ECBM), which were tested and discussed in the literature mainly in the 1990’s and 2000’s (e.g. Reeves 2001; van Bergen et al. 2006). The enhanced methane recovery process in shales is different from the volumetric displacement of methane in coal reservoirs, in which a higher affinity for CO2 compared to CH4 drives the methane recovery and results in high amounts of CO2 stored in the rock. Challenges are the commercial availability of CO2, kinetics of CO2/CH4 exchange and swelling of the coal matrix due to CO2 sorption that will significantly reduce permeability.
Applying these concepts to shale gas reservoirs, the potential benefits remain largely unchanged, while the drawbacks and complications will be amplified:
As for coal, the CO2 sorption capacity of clays seems to be higher compared to CH4 (e.g. Chareonsuppanimit et al. 2012; Weniger et al. 2010), but the overall sorptive uptake significantly lower. This is because high sorption capacities in coal are linked to its high organic matter content, which is much lower in shale. Replacing CH4 by CO2 in shale is largely a diffusion-driven process. Fracture spacing in shale determines matrix block sizes, with larger fracture spacing resulting in lower CO2 accessibility. Permeability in coal is orders of magnitude higher (mDarcy) compared to shales (μ to nDarcy), indicating a much more narrow cleat network in coal compared to fracture networks in shale, or larger cleat compared to fracture apertures. Large fracture spacings require long diffusive travel times, hence slow exchange rates for the sorbed gas.
Coal shrinks with CH4 desorption and swells with CO2 adsorption. Shrinkage and swelling are directly related to the amounts of sorbed gas. Because sorption capacity for CO2 is larger than for CH4, coal swelling is more pronounced with CO2 adsorption, and leads to permeability reductions by orders of magnitude. Shale swelling with CO2 sorption has not been verified, but is plausible (as discussed above), and might cause similar issues. Gas shale reservoirs typically require reservoir temperatures that passed the gas window at temperatures around or higher than 100 °C. This might have caused a significant or even complete transformation of swelling clays to non-swelling species, as has been discussed above.
In summary, we consider the utilisation of shale reservoirs or shale formations for storing CO2 or for an enhancement of natural gas recovery challenging and not well understood. A technology that did not reach a commercial stage when applied to coal beds seems to not benefit from lower permeabilities, lower sorption capacities and generally larger depths in shale gas formations. For sure, more research is needed to better understand the interplay between all these parameters.
We discussed the implications of CO2 clay mineral sorption and swelling and potential consequences for CO2 storage containment and overall trapping potential in storage reservoirs. While we aimed at a general overview of the topic we realise that this research field is still quite immature and requires further work. This is especially true for formations with high smectite contents. We attempted to raise a number of critical issues related to CO2-clay and CO2-caprock interactions with the goals of creating awareness and initiating discussions on CO2-clay interactions that go beyond the laboratory scale. Some of the fundamentals have been addressed in recent work of the authors of this paper or by other researchers. The transfer of these fundamentals into geological applications is recommended for the future and this transfer should mainly address the geological risks of storing CO2 underneath caprocks with high contents of swelling clays.
The specific aspects addressed here can be summarised as follows: For reservoir rocks we find that clay minerals (smectite, illite, kaolinite) can act as a sink for carbon storage by physical adsorption of CO2. Clays in general have a high specific surface area and it was shown that the sorption capacity correlates well with the supermicropore/mesopores/macropore surface areas determined using N2 low pressure sorption. The sorptive trapping certainly depends on the overall clay content in the reservoir formation. Considering however a fast reaction rate and potentially comparable trapping capacities in the form of dissolution or mineral trapping, it should be considered in the evaluation of certain storage projects. In addition to reservoir rocks we can state that for intact caprocks, fluid transport through the matrix occurs by diffusion only, and little or no leakage is expected. Diffusion-driven sorption increases concentration gradients from the base to the top of the sealing formation. Therefore a slight increase in diffusive fluxes can be expected. Nevertheless, diffusive transport is slow and diffusive leakage is probably irrelevant over time scales of at least thousands of years. Some care should be taken when a dry or near dry CO2-plume gets in contact with the reservoir/seal interface by buoyancy. Smectite bearing shale could dewater by pore or interlayer water dissolution in CO2. As a consequence dehydration cracks are plausible; their frequency and propagation into the seal formation depends on diffusion coefficients, plume saturation, seal porosity etc. Another potential leakage mechanism relates to wellbore annuli that can develop between cement and host rock following thermal effects or non-perfect cementation. If dry CO2 is migrating upwards along such an annulus, pore water in the seal will be dissolved and transported to more shallow reservoirs, or to the surface. Similar to the situation for intact seals, this process could lead to dehydration cracks, possibly increasing the leak rates along the well. In contrast to intact seals or wells we expect (partial) healing of fractures subject to normal stresses. This is because smectite aligned at the fracture surfaces will swell and lead to decreasing fracture apertures and flux rates. In contrast it was found that clay swelling may result in the development of swelling stresses that under certain circumstances, can potentially lead to shear-type failure. As a consequence faults may get activated if impact area is large enough. However shear-type failure does not necessarily lead to fluid leakage. This depends on many factors, among others the contact area, mineralogy in the fault zone, effective normal stress or reservoir pressurization. Detailed case-by-case investigations are needed for risk assessment of storage sites that are transected by faults.
This research has been carried out in the context of the TKI Toeslag 2013 project, part of the CATO-2-program (www.co2-cato.org). CATO-2 is the Dutch national research program on CO2 Capture and Storage technology (CCS). The program is financially supported by the Dutch government (Ministry of Economic Affairs) and the CATO-2 consortium parties. G. R. was supported as part of the Nanoscale Control of Geologic CO2 (NCGC) Center, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences.
- Amann A, Weniger P, Krooß BM, Littke R, Bertier P, Stanjek H, Draeger I, Clauser C, Blume J, Eckhart J-D, Stosch HG, Neumann T, Mutschler T, Balthasar K, Triantafyllidis T (2011) CO2 seals—integrity of sealing rock formations for CO2 storage. BMBF/DFG special program, GeotechnologienGoogle Scholar
- Audigane P, Gaus I, Xu T, Pruess K (2005) Reactive transport modeling using TOUGHREACT for the long term CO2 storage at Sleipner, North Sea. In: Fourth annual conference on carbon capture and sequestration, Alexandria, VA, p 17Google Scholar
- Bergaya F, Theng BKG, Lagaly G (2006) Handbook of clay science. Dev Clay Sci 1:1224Google Scholar
- Bertier P, Schweinar K, Stanjek H, Ghanizadeh A, Clarkson CR, Busch A, Kampman N, Prinz D, Amann-Hildenbrand A, Krooss BM, Pipich V, Di Z (in press) On the use and abuse of N2 physisorption for the characterization of the pore structure of shales. Clay Clay MinerGoogle Scholar
- Bois A-P, Garnier A, Rodot F, Saint-Marc J, Aimard N (2009) How to prevent loss of zonal isolation through a comprehensive analysis of micro-annulus formation. In: SPE annual conference and exhibition (SPE 124719), New OrleansGoogle Scholar
- Brigatti MF, Galan E, Theng BKG (2013) Chapter 2—structure and mineralogy of clay minerals. In: Bergaya F, Lagaly G (eds) Developments in clay science. Elsevier, Amsterdam, pp 21–81Google Scholar
- Busch A, Bertier P, Gensterblum Y, Giesting P, Guggenheim S, van Groos AK, Weniger P (2012) Clay/CO2 interactions in the context of geological storage of carbon dioxide, 3rd EAGE Shale workshop-shale physics and shale chemistryGoogle Scholar
- Fjær E, Holt RM, Horsrud P, Raaen AM, Risnes R (2008) Petroleum related rock mechanics, 2nd ednGoogle Scholar
- Iglauer S, Wülling W, Pentland CH, Mansoori SKA, Blunt MJ (2009) Capillary trapping capacity of rocks and sandpacks, EUROPEC/EAGE conference and exhibition. Society of Petroleum Engineers, AmsterdamGoogle Scholar
- Kang SM, Fathi E, Ambrose RJ, Akkutlu IY, Sigal RF (2010) CO2 storage capacity of organic-rich shales. In: SPE annual technical conference and exhibition. Society of Petroleum Engineers, FlorenceGoogle Scholar
- Krooss BM, Leythaeuser D, Schaefer RG (1992) The quantification of diffusive hydrocarbon losses through cap rocks of natural gas reservoirs—a reevaluation. AAPG Bull 76:403–406Google Scholar
- Loizzo M, Akemu OA, Jammes L, Desroches J, Lombardi S, Annunziatellis A (2011) Quantifying the risk of CO2 leakage through wellbores. Society of Petroleum Engineers SPE-139635-PAGoogle Scholar
- Meunier A (2005) Clays. Springer, HeidelbergGoogle Scholar
- Nuttall BC, Drahovzal JA, Eble CF, Bustin RM (2009) Regional assessment of suitability of organic-rich shales for carbon sequestration: an example from the devonian shales of the Illinois and Appalachian Basins, Kentucky. In: Grobe M, Pashin JC, Dodge RL (eds) Carbon dioxide sequestration in geological media-state of science. AAPG Stud Geol, pp 173–190Google Scholar
- Reeves S (2001) Geological sequestration of CO2 in deep, unmineable coalbeds: an integrated research and commerical-scale field demonstration project. SPE J 71749:1–11Google Scholar
- Reiss LH (1980) The reservoir engineering aspects of fractures reservoirs. Institut Francaise du PetroleGoogle Scholar
- Rother G, Ilton ES, Wallacher D, Hauss T, Schaef HT, Qafoku O, Rosso KM, Felmy AR, Krukowski EG, Stack AG, Grimm N, Bodnar RJ (2013a) CO2 sorption to subsingle hydration layer montmorillonite clay studied by excess sorption and neutron diffraction measurements. Environ Sci Technol 47:205–211CrossRefGoogle Scholar
- Snippe J, Wei L, Tucker O (2012) Containment impact of calcite pathways in the primary caprock of CO2 storage in a depleted North Sea gas field. In 11th greenhouse gas technology conference, KyotoGoogle Scholar
- Spiering P, Wentinck R, Busch A, Fraaije JGEM, Tambach TJ (2014) Molecular mechanisms of CO2 in swelling clays, Goldschmidt 2014, SacramentoGoogle Scholar
- Thommes M, Kaneko K, Neimark Alexander V, Olivier James P, Rodriguez-Reinoso F, Rouquerol J, Sing Kenneth SW (2015) Physisorption of gases, with special reference to the evaluation of surface area and pore size distribution (IUPAC Technical Report), Pure and Applied Chemistry, p 1051Google Scholar
- Velde B, Vasseur G (1992) Estimation of the diagenetic smectite top illite trans-formation in time-temperature space. Am Mineral 77:967–976Google Scholar
- Wang Y, Papamichos E, Dusseault MB (1996) Thermal stresses and borehole stability. A.A. Balkema. Permission to Distribute—American Rock Mechanics AssociationGoogle Scholar
- Wentinck HM, Busch A (2014) The swelling of smectites by high-pressure CO2 and the possible impact on the containment of CO2 below the caprock, Mid European clay conference, DresdenGoogle Scholar
- Zhang M, Spiers CJ, Busch A, Wentinck R (2014) Swelling stress development in smectites upon exposure to CO2 under in situ P-T conditions. In: Mid European clay conference, DresdenGoogle Scholar