Micro- and Macroscale Consequences of Interactions Between CO 2 and Shale Rocks

: In carbon storage activities, and in shale oil and gas extraction (SOGE) with carbon dioxide (CO 2 ) stimulation fluid, CO 2 comes into contact with shale rock and its pore fluid. As a reactive fluid, the injected CO 2 displays a large potential to modify the shale’s chemical, physical, and mechanical properties, which need to be well studied and documented. The state of the art on shale–CO 2 interactions published in several review articles does not exhaust all aspects of these interactions, such as changes in the mechanical, petrophysical, or petrochemical properties of shales. This review paper presents a characterization of shale rocks and reviews their possible interaction mechanisms with different phases of CO 2 . The effects of these interactions on petrophysical, chemical and mechanical properties are highlighted. In addition, a novel experimental approach is presented, developed and used by our team to investigate mechanical properties by exposing shale to different saturation fluids under controlled temperatures and pressures, without modifying the test exposure conditions prior to mechanical and acoustic measurements. This paper also underlines the major knowledge gaps that need to be filled in order to improve the safety and efficiency of SOGE and CO 2 storage.


Introduction
Shale rocks come in contact with carbon dioxide (CO2) when the latter is pumped underground either for CO2 capture and storage (CCS) or for shale oil and gas extraction (SOGE) purposes [1]. In the case of CCS, the shale typically acts as a caprock (sealing) that protects the CO2 stored in a reservoir underneath from escaping back to the atmosphere. In SOGE, the CO2 is used to enhance natural gas production. It can be used either during the stimulation stage as a fracturing fluid or at the secondary gas recovery stage as an extraction fluid displacing the natural gas and increasing the extraction efficiency. In this case, the CO2 is pumped directly into the shale formations (using shale as a reservoir).

Shale as a Caprock
Since approximately 1950, the concentration of carbon dioxide (CO2) in the atmosphere has increased abnormally. It is widely believed that this increase is one of the main causes of current global warming [2][3][4]. One of the solutions to combat this increase is to capture anthropologically emitted CO2 and store it in suitable reservoirs. Among the CO2 capture and storage (CCS) alternatives, geological sequestration in high porosity reservoirs is currently the most mature technology to efficiently combat climate change. The feasibility of CCS has been demonstrated in

Shale Rock Characteristics
Shale is a laminated or fissile [26] fine-grained detrital sedimentary rock, formed by the consolidation of clay, silt, or mud ( Figure 1) [27]; it is also the most abundant sedimentary rock on Earth [28][29][30]. In rock mechanics, shale can be defined as a rock where clay minerals form the loadbearing framework [31][32][33]; however, strictly speaking, freshly deposited clays are not shale. The lithological properties of shale (e.g., the textural properties, elevated capillary entry pressure, and ductility) are mostly controlled by the shale's mineralogical composition [34]. Moreover, shales are highly inhomogeneous and anisotropic due to their fine laminations and the constitutions of different clay minerals.

Mineralogical Properties of Shales.
The complex and wide variety of composition types among shales can be attributed to their different sources (i.e., different rocks, reliefs, and climate), the degree of weathering (decomposition from the source rock and during transportation), the end-product of weathering and chemical interactions, and biochemical additions [26,35]. Moreover, the alteration of some clay minerals occurs due to chemical diagenesis with an increase in burial depth, creating greater complexity in shale composition. The composition of shale is dominated by clay minerals, which are mostly hydrated aluminium silicate with some replacement by iron and magnesium. Typical minerals in shale include (e.g., [26]) the following: Other minerals include biochemical carbonates (e.g., calcite, dolomite) along with iron-bearing minerals (e.g., pyrite, siderite, and hematite). The presence of organic carbon makes certain shale formations potential petroleum (oil/gas) sources and/or reservoirs (gas shale). The exchange of cations in shales commonly occurs in clay minerals, which can be classified by their layered lattice structures. Two-layer (1:1) lattice structures contain one tetrahedral and one octahedral layer (Figure 2Error! Reference source not found.), whereas three-layer (2:1) lattice structures are made of one octahedral surrounded by two tetrahedral layers ( Figure 3). These layer units are linked together by water or cations. The layer charge depends upon the substitution of cations in the tetrahedral or octahedral sheets [36]. Kaolinite group minerals generally have no (or very little) layer charge, whereas illite typically has a layer charge of less than 1 (0.7-0.9 per O10(OH)4). Smectite possesses a layer charge between 0.2 and 0.6 per O10(OH)4. Kaolinite is known as a neutral mineral, and its interlayer spaces are tightly bounded by hydrogen bonds, extending from exposed hydroxyl ions in the octahedral sheet of one layer to the oxygen layers in the tetrahedral sheet of the next layer. Almost no ionic replacement occurs at either the tetrahedral or the octahedral sheets in kaolinite.  In illite, the interlayer space is also bonded by two layers of water and large K + ions, which are not easy to substitute. On the other hand, the interlayer space in smectite generally contains two layers of water and exchangeable cations (Na + , Ca +2 ). Due to the low layer charge in smectite, the smectite expands when it comes in contact with fluids with higher fluid activities and is substituted by the cation of preference (e.g., Ca +2 can be replaced by Na + ). Smectite lattices shrink when native Na+ is replaced by K+ in its interlayer space.
Therefore, the cation exchange capacity depends on the degree of layer charge and chemical ability to exchange the cation in the interlayer space by foreign cations. The cation exchange capacity (CEC) of clay minerals ranges from 1-10 mequ/100 g for kaolinite to 80-150 for smectites and 120-200 for vermiculites [39]. Depending on the mineralogical composition of shale, its CEC values can differ significantly. However, cationic substitution takes place not only in the interlayer space but also at the tetrahedral or octahedral layer's lattice. The transformation of clay minerals depends on the cationic transformation/exchange in the layer's lattice, which appears to occur with exposure to appropriate conditions (e.g., pressure, temperature, and presence of a solution). As mentioned, shales also contain organic carbon. Most shales contain a small percentage (typically less than 10%) of organic matter. Organic matter is a source for a higher porous area within low porosity shales, but this porosity depends on the maturity of the organic matter.

The Petrophysical and Mechanical Characteristics of Shale
The petrophysical properties of shale include, among others porosity, permeability, grain size and shapes, and a specific surface area. The two most important parameters are porosity and permeability. Pore space (intrinsic void or fractures) are the storage sites for gas/liquid. The porosity of freshly deposited mud can be as high as 70% [40] and can be reduced to about 30% beyond a few hundred meters depth [41] and as low as below 10% at larger depths [42,43]. The porosity of shale depends on the depth of its burial (compaction), as well as its mineralogical composition, texture, and degree of diagenesis. The connective pores, along with existing fractures, create the pathway for fluid migration and thus contribute to the permeability of the rock. The physical and chemical interactions of fluids with shale strongly depend on the shale's porosity and permeability. Therefore, it is worthwhile to understand permeability when analysing any effect of fluid (such as CO2) on argillaceous rocks, such as shale. Unfortunately, it is very difficult to achieve reliable permeability measurements for shales, given their fine pore sizes and complex structures [35,44,45]. In addition, the permeability relative to different fluids (gas or liquids) is different for the same rock sample. Shales exhibit very low hydraulic permeability, and this value varies widely, ranging from nanoDarcy (nD) to microDarcy (µD) [46]. The equation derived by Kozneny [47] and later modified by Carman [48] indicates that permeability is a function of porosity, fluid viscosity, grain size and shape, tortuosity, the pressure gradient across a cross-section, and the specific surface area. However, for CO2 storage integrity, scale-dependent permeability may be worthy of note. Many laboratoryderived permeability measurements underestimate the large (reservoir) scale permeability [49], which can be attributed to the existence of fractures and non-clay minerals [45]. Scale dependency has also been reported by several authors (e.g., Bredehoeft et al. [50], Rudolph et al. [51], and Keller et al. [52]). However, Neuzil [45] showed that the permeability scale-dependence in argillaceous rocks is not present at an intermediate scale, but it may be present at very large regional scale. It is also important to study the fracture network of the sealing material since there is field evidence (e.g., [53]) of fluid movement through fracture networks, which may even lead fluids all the way up to surface (e.g., Bond et al. [54], Ingram and Urai [55], and Lewicki et al. [56]).
A few more physical or mechanical characteristics of shales are worthy of mention, such as capillary entry pressure, strength, stiffness, acoustic velocity and anisotropy [57] (acoustic and strength), wettability, etc. As mentioned earlier, capillary entry pressure is one of the most important properties to ensure that shale is sealed and to quantify the maximum height of the injected CO2 column that can be held in the reservoir [34,58,59]. A higher capillary entry pressure of shale restricts the movement of the fluid (e.g., CO2), which in turn restricts the available space for the fluid to react with the shale [59]. The potential leakage of CO2 through caprock could occur by diffusion, capillary breakthroughs, or by fracture flow; among these mechanisms, the latter two dominate [60][61][62]. These seepages could create potential available space for CO2-shale interactions inside the shale formation (the seal), in addition to the contact area between the seal and the reservoir. The capillary entry pressure is directly related to the permeability of the shale material, the interfacial tension between the shale wetting fluid and the non-wetting fluid, and the cation exchange capacity [63]. The very low permeability, very small pore throat radius, and higher cation exchange capacity of the shale contribute to a very high capillary entry pressure for any non-wetting fluid to enter into the generally water-wet shale [55,63,64]. The fluid-flow and fluid-recovery efficiency in shale reservoirs can also be affected by shale wettability [65][66][67]. However, the wettability of shale is still ambiguous, ranging from oil-wet to water-wet [66]. This can be modified by the interplay of various factors, such as pH, temperature, and surface access [67]. Since fine-grained sediments (mud/shale) are deposited in a marine environment, the shale is expected to be water-wet [68]. However, the presence of mature organic matter may cause local oil-wet patches [69].
Among the many characterization parameters, stiffness, strength, and acoustic velocities will be briefly discussed in this review article. These parameters are generally related to the porosity, density, mineralogy (mainly clay content), and heterogeneity of shale materials. Stiffness is a very important parameter in geomechanics to understand the deformability of rock and can be obtained from static (stress-strain) or dynamic (acoustic velocities) measurements. The stiffness parameter can bridge the geomechanical parameters and seismic parameters (velocity). The number of stiffness parameters needed to define a rock depends on the rock's anisotropy. For an isotropic linear elastic material, only two parameters are required to characterize the material: Young's modulus and Poisson's ratio or the shear and bulk moduli. However, when the material is anisotropic, the number of stiffness parameters increases significantly (to a maximum of 81). However, shale can be approximated as a transversely isotropic (TI) material where the properties of the rock are similar in two horizontal axes (x-and y-axis) but different from its vertical axis (z-axis) (e.g., x = y ≠ z). The number of the stiffness parameter in the case of TI is 5 (for more details, see Fjaer et al. [31]). Very few articles have reported the effects of CO2 on stiffness parameters. Both the Young's modulus and Poisson's ratio were investigated by Agofack et al. [70] and Espinoza et al. [71], but only changes in Young's modulus was reported by Lyu et al. [72]. More detailed discussions on CO2's effect on shale stiffness are presented in Section 5.2.
The strength of the shale is very important, not only to understand how difficult the shale is to break but also for seal integrity evaluation, well planning, wellbore stability, reservoir compaction, and surface or sea floor subsidence, which has been found to be sensitive to the internal properties of rock or external factors such as composition, organic content, pore pressure, and stress history (e.g., Dewhurst et al. [42]). The strength of shales varies with the direction of measurement (anisotropy), e.g., shales are the strongest at 0° with the bedding direction (measuring the strength across the bedding) and are the weakest at 45-60° with the bedding direction. Strength parameters include UCS (unconfined compressive strength), shear strength, and tensile strength. For more discussions, with a few examples of the effects of CO2 on shale strength, please see Section 0.
Anisotropy is one of the most widely studied properties (yet to be understood perfectly) of shale. When a rock property (e.g., strength, acoustic velocity, or permeability) varies with the direction of measurement (angle of measurement) with respect to a fixed orientation (e.g., bedding), the sample is said to be anisotropic in that property (e.g., it can have velocity anisotropy or fracture anisotropy). Shales show two types of anisotropy: intrinsic (lithological anisotropy, developed from the preferred orientation of platy clay minerals and the development of thin lamination) and stress-induced anisotropy (anisotropy developed with an increase in stress). There are many articles in the literature discussing strength anisotropy (e.g., Jin et al. [73], Fjaer and Nes [74]), where the variation of the strength of the rocks with the angle creates problems, especially in inclined wells (e.g., borehole stability). Few researchers have discussed the effects of stiffness, strength, acoustic velocity, and even fracture anisotropy on CO2 storage or seal integrity. Bond et al. [54] discussed the influence of fracture anisotropy on CO2 flow. Cheng et al. [75] and Armitage et al. [76] investigated the effects of permeability anisotropy on buoyancy driven CO2 flow. Al Ismail et al. [77] investigated the effects of CO2 adsorption on permeability anisotropy. Taheri et al. [78] made an attempt to investigate the effects of anisotropy and heterogeneity in both the horizontal and vertical directions of layering on the CO2 dissolution in a saturated porous medium with brine using simulation methods. Lu et al. [79] reported anisotropic strain in response to CO2 injection, where the strain is always smaller in the direction parallel to the bedding plane.

Water Sensitivity of Shale Rocks
The main cause of the water sensitivity of shale formations is the presence of swelling clays in their composition. The most common type of swelling clay is smectite. Smectite's structure consists of 2:1 layers consisting of an octahedral hydroxide sheet sandwiched between two opposing tetrahedral silicate sheets [80]. When the fluid molecules are adsorbed into the interlayer, the interlayer distance d001 increases and induces swelling of the material. Ferrage [80]  where Å refers to Ångström (1 Å = 0.1 nanometer). The exposure of shale to fluids with compositions even slightly different from those of in situ pore fluid induces a difference in chemical potential and changes the swelling pressure [81]. Under constant strain conditions, the increase of pressure due to adsorption-induced swelling can be as high as a hundred MPa up to a few GPa [82,83]. If the swelling pressure along with the pore pressure exceed the in-situ stress and particle cementation forces, shale failure will occur.

Processes at the CO2/Brine Interface
When dry CO2 is pumped underground, it comes into contact with the reservoir brine. Two processes take place at the brine/CO2 interface: (1) Water present in the brine evaporates to the CO2 phase [84], and the concentration of salts in the brine increases. With the progressive evaporation of water, the brine becomes oversaturated with salts, and salt precipitation may occur. (2) CO2 dissolves in brine. The dissolved CO2 reacts with water to form carbonic acid, which partially dissociates and forms carbonic (CO3 2− ), bicarbinate (HCO3 − ), and hydrongen (H + ) ions. Due to these dissolution and dissociation processes, the pH of CO2-rich brine drops. The pH value at equilibrium depends on the partial pressure of CO2 and on the temperature and salinity of the brine [85][86][87]. These acidic conditions of the pore fluid have the potential to affect the composition of the rock matrix. In addition to CO2-rich brine, dry CO2 can also affect the rock microstructure and its mechanical properties via different mechanisms. These interaction mechanisms are described below.

Drying
At the dry CO2/brine interface, water evaporates and mixes with the CO2 phase. Under downhole conditions, this evaporation and mixing lasts until the CO2 becomes saturated with water (i.e., reaches equilibrium). The thermodynamics of water partitioning in equilibrium with CO2 have been described by Spycher et al. [86,88]. The water absorption into CO2 leads to drying of the surrounding formation fluids [89][90][91][92][93][94]. This drying may affect both the brine present in the pore matrix and the rock matrix itself. The continueous drying of brine, filling pore spaces in the rock matrix, often leads to the oversaturation and precipitation of different minerals in the matrix. It has been found that drying and concommitant salt precipitation may lead to the clogging of an initially very permeable pore matrix, which may eventually result in injectivity problems [89,95,96]. The permeability of nonfractured shale rocks is usually very low, but, as fractures are very abundant in shale rocks, the overal permeability of fractured shales can be significantly affected by the drying and concommitant salt precipitation processes. More specifically, fractures serving as flow conduits may be clogged by the precipitating salts. Nooraiepour et al. [97] studied how fractures in Draupne shale (primary caprocks for Smeaheia CO2 storage in Norway) clog upon CO2 induced salt precipitation. The authors showed that the CO2 phase influences the magnitude, distribution, and precipitation patterns of the salt in fractures. The injection of gaseous CO2 resulted in larger salt precipitation compared to both the liquid and supercritical CO2. Miri et al. [92,93] demonstrated the importance of capillary forces on salt precipitation. It has been suggested that drying-induced salt precipitation may have fracture sealing potential [97]. However, it remains unknown whether salt precipitation has any effect on the mechanical properties of shale rocks close to fractures and to to what extent it influences fracture propagation.
Not only interstitial brine but also the rock matrix itself may be affected by drying. Drying may cause rock dessication. It is known that shale dehydration will lead to structural changes like shrinkage and cracking within the shale matrix [24]. This occurs as a consequence of the dehydration of clays, which tends to respond, under volumetric changes (swelling or shrinkage), to changes in brine/water content (wetting and drying) [25]. The integrity and strength of Pierre shales were found to be deterioted by submitting the samples to successive wetting and drying cycles [26,27]. Mugridge and Young [98] observed significant deterioration of mechanical properties, despite negligable mineralogical changes. A reduction in size of the clay aggregates was suggested to contribute to the observed residual strength decrease [27]. The above-described observations were made under ambient conditions. The effect of dry CO2 on shale dessication and the possibly-induced changes in the mechanical properties of shales under realistic downhole stress conditions seem to be neglected in the literature and are perceived by the authors as a knowledge gap that needs to be filled in order to obtain a better understanding of the interactions between dry CO2 and shale reservoirs or caprocks. This knowledge gap has already been highlighted by Gaus [28], who suggested using mass balance calculations to answer the question of whether, and under which CO2 injection conditions, the dessication of the caprock may impair its integrity.

Dissolution of Organic Matter
Another physical process that may affect the microstucture, and thus the mechanical properties, of shale is the dissolution of organic matter (i.e., kerogen) present in the shale matrix. Kerogen is the major organic component of shale rocks and consists of a complex mixture of organic materials. Under a supercritical state (a temperature and pressure higher than 31.1 °C and 7.2 MPa, respectively), CO2 has been demonstrated to be an excellent solvent for organic compounds [99]. Supercritical CO2 has a density and dissolving ability similar to those of liquid CO2, while its diffusivity, viscosity, and surface tension are comparable to those of gas [100]. As such, supercritical CO2 offers the efficient dissolution and mass transfer of dissolved components [101]. CO2 is a nonpolar solvent and thus it dissolves readily nonpolar molecules [102]. The dissolution of polar oil components in supercritical CO2 was, however, enhanced in the presence of polar solvent additives like methanol [102]. In addition, it has been shown that supercritical CO2 can partially dissolve the kerogen present in shale rocks [100,103]. The supercritical extraction efficiency of kerogen was found to be related to the mineralogical composition and microstructure of shale rocks and was higher for carbonate rich shales [102]. The corelation between kerogen dissolution and changes in the porosity, permeability, and mechanical properties of shale rocks has not yet been well established. It is thus unclear how the long-term exposure of organic matter-rich shales to supercritical CO2 can affect their mechanical stability.

Sorption of CO2 on Shale Rocks
This phenomenon has been extensively studied both in the context of enhanced oil and gas recovery and in the context of CO2 sequestration [20,[104][105][106][107][108][109][110]. Heller and Zoback [111] compared the sorption capacity of CO2 and that of methane on clay minerals (kaolinite and illite) and carbon as a model for shale components. The authors found that the sorption capacity of CO2 on both minerals and organic matter was 2-3 times higher compared to methane. According to Nuttall et al. [112] and Kang et al. [105], the difference in the sorption capacity between CO2 and methane can be up to 5-10 times depending on the shale type. Upon the adsorption and absorption of CO2, shales undergo volumetric changes [79,111], which are proportional to the amount of adsorbed CO2. For a given amount of adsorbed gas, the magnitude of swelling observed for different shales and shale components is different. The difference in swelling capacity has been attributed to the difference in the stiffness between materials (i.e., the stiffer the material was, the smaller the volume change that was found) [111]. The complexity of the swelling mechanisms of shales upon the adsorption of CO2 arises from the chemical diversity of shale components. Shale components can adsorb and absorb CO2 in both physical and chemical manners known as physi-and chemisorption. Figure 4 schematically illustrates the adsorption of CO2 at mineral grain surfaces, as well as its absorption inside organic matter and clay mineral grains under true dry conditions. The deformation of clay particles and organic matter upon the adsorption of CO2 is probably the main contributor to global shale volumetric changes when the shale is exposed to CO2 [111,113,114]. It has been shown that the spacing between the crystalline structure of clay sheets increases upon adsorption of CO2 [115]. For example, for a single hydrated Na-montmorillonite (with original spacing of d001~12.4 Å), the spacing increaed by 0.5 Å, which resulted in a 4% expansion in the direction perpendicular to the clay sheets [116]. The expansion determined for Na montmorillonite clays initially containing a higher fraction of water in the interlayer region has been reported to reach 9% [117]. The magnitude of this so called "crystalline swelling" is also dependent on the type of cation present in the interlayer [115]. Therefore, it is expected that the magnitude of clay swelling inside the shale matrix will be a result of CO2 adsorption and possible concomitant ion exchange processes. Molecular dynamic simulations show that the organic matter present in shale rocks (kerogen) undergoes non-negligable deformation upon CO2 uptake [113,118]. Ho et al. [118] suggest that strong chemo-mechanical coupling exists between the gas sorption and mechanical strain of a kerogen matrix. The authors showed that the kerogen volume can expand by up to 11% upon CO2 adsorption at around 19 MPa and that gas uptake is enhanced by kerogen swelling [118]. It has also been suggested that the CO2-induced swelling of clays and kerogen may contribute to the sealing of small fractures and thus to a reduction in the permeability and improved sealing capacity of swelling clayrich shale rocks [117,119]. However, this hypothesis needs to be experimentally validated. The magnitude of net shale swelling will depend on the ratio between the total content of swelling matter (i.e., the swelling clays and kerogen) and nonswelling matter (e.g., quartz and nonswelling clays) in the shale matrix. The effect of CO2 sorption-induced swelling on the mechanical properties of shale is, however, unknown. The main difficulty remaining is to decouple sorption from other physicochemical processess (such as the dissolution of minerals) that occur within shales during their exposure to CO2 or from the processes of pH-induced swelling.

Dissolution of Minerals
When shale comes in contact with CO2-saturated brine or wet CO2, the carbonic acid present in the solution starts diffusing into the shale's pore space. In shale's composition, there are always some minerals susceptible to carbonic acid's effects. These minerals may be dissolved and eventually leach out. Shales may have very diverse mineralogical compositions (see Section § 0), but they are most often comprised of quartz, feldspar, clay minerals, carbonates, iron oxides, and other minerals, as well as organic matter (kerogen) [120]. All these minerals can, to some extent, react with acid under very different dissolution rates.
Poorly water soluble carbonates like calcite (CaCO3), dolomite (CaMg(CO3)2), magnesite (MgCO3), and siderite (FeCO3) are very susceptible to acid attack and react rapidly with carbonic acid according to Equation (1): MeCO H Me HCO As a consequence of this reaction, carbonate is dissolved, protons are consumed, pH increases, and both metal ions (Me 2+ ) and bicarbonate ions (HCO3 − ) are released into the solution. Lahann et al. [121] showed that carbonate dissolution was one of the first responses resulting from the injection of CO2 into the New Albany Shale reservoir [121]. The dissolution rates of calcite, dolomite, and magnesite in CO2-saturated aqueous solutions at partial CO2 pressure up to 5.5 MPa (pH between 3 and 6) have been well described by Pokrovsky et al. [122] and Golubev et al. [123]. The dissolution rate was found to decrease in the following order: calcite > dolomite > siderite > magnesite [123]. As the carbonate dissolution rates are relatively fast at pH conditions relevant for storage and fracturing, the reactivity of low permeability shales will be, in most cases, transport-controlled. The dissolution and reaction of CO2 with water, the dissociation of acid, and the diffusion of the dissociated acid to a carbonate substrate are the limiting factors in carbonate leaching [87]. Therefore, any increase in the porosity and permeability of the shale matrix will facilitate carbonate leaching [124]. Carbonate leaching from shale rocks upon exposure to CO2-saturated brine has been reported by many authors [124][125][126][127]. Armitage et al. [128] exposed "In Salah" caprock samples to CO2 saturated brine and showed that the caprock experienced a significant increase in porosity and permeability due to the dissolution of its siderite and chlorite minerals. More interestingly, Sanguinito et al. [129] observed the carbonate dissolution in Utica shale upon exposure to dry CO2. They showed that the presence of interstitial pore water is enough for carbonation processes to occur in shale. Recently, Goodman et al. [130] showed that, by comparing unexposed exposed outcrop Utica Shale samples, both the porosity and the surface area (m 2 /cm 3 ) significantly increased when exposed to CO2 and that this effect is higher when exposed to a H2O + CO2 solution. Performing relatively similar experiments on Marcellus Shale samples, Kutchko et al. [131] showed that exposure to CO2 decreases the BET surface area and porosity of a downhole sample; the effect on the outcrop sample was also significant, but sometimes showed an opposite trend. Thus, the physicochemical and mechanical changes in shale rocks saturated with brine exposed to both wet and dry CO2 should continue to be addressed.
Phyllosilicate minerals are also susceptible to carbonic acid attack. They present a wide range of chemical constituents, including smectites, kaolinite, serpentine, talc, mica, illite, and chlorite minerals, but the common feature for all phyllosilicates is their layered nature. Due to this sheeted structure, their specific surface area is large, which increases their reactivity. In most cases, the dissolution of phyllosilicates is sensitive to pH and increases with a decreasing pH in an acidic environment [87]. Therefore, reactivity increases with an increase in partial CO2 pressure. The reaction of chlorite, as an example of a phyllosilicate mineral, in the presence of water and carbonic acid is described in Equation (2). Due to the complexity of the structures and pathways by which the dissolution of chlorite and other phyllosilicates may proceed, only a generalized reaction is presented [132,133] Chlorite dissolution releases silica, iron, and/or magnesium and aluminate ions into the solution. In contrast with other phyllosilicates, the chlorite dissolution rates were found to be independent of pH, with a pH range between 4.2 and 5.3 [132]. Feldspar undergoes hydrolysis, as outlined in Equation (3) The substrates of K-feldspar hydrolysis in the presence of carbonic acid are potassium carbonate, silica dissolved in water, and solid kaolinite [134]. Thus, during the hydrolysis reaction, soluble elements (potassium carbonate and silica) are leached out of the rock; only the insoluble kaolinite is left behind. Similar to the reaction with carbonates, feldspar hydrolysis leads to the consumption of acid and the neutralization of pH. The long-term reactive transport modelling of CO2 in the Nordland Shale caprock at Sleipner, performed by Gaus et al. [135], shows that carbonate dissolution initially dominates, but, in the long term, feldspar alteration is the dominant reaction.
In contrast to carbonates and silicates, the dissolution of quartz is a slow process under CO2 storage or fracturing conditions. According to Knauss and Wolery [136], the dissolution rates of quartz are independent of pH for pH values lower than 6. This means that increasing CO2 pressure and thus decreasing pH will not facilitate quartz dissolution [87]. The relatively small quartz grain surface area compared to, for example, feldspar can further slowdown quartz's dissolution in CO2saturated brine. Consequently, the effect of CO2-saturated brine on quartz is usually assumed negligible.

Precipitation and Reprecipitation of Minerals
The reactions of carbonates and other minerals with CO2-saturated brine leads to the consumption of proton and thus increases the pH of brine. Along with these pH changes, metal ions are released to the solution, as described in reactions (1)-(3). At relatively high concentrations of metal ions, this increase of pH may lead to the reprecipitation of some minerals [135,137]. The reprecipitation of dolomite, ankerite, siderite, and iron hydroxide was experimentally shown to take place in shales exposed to dry CO2 or CO2-saturated brine [121,129,138]. Kampman et al. [138] examined caprock samples recovered from a natural CO2 reservoir. This reservoir was located at a shallower depth and had a lower pressure than that encountered in reservoirs for geological carbon storage. However, since the reactions between the minerals and CO2-saturated brines are only slightly affected by pressure, the authors were able to assess the long-term impacts of CO2-rich fluids on caprock integrity. The authors showed that the caprock, at its interface with the reservoir (up to 7 cm), was depleted of hematite and primary dolomite and that the minerals were reprecipitated in the form of ankerite-dolomite, pyrite, and gypsum. The authors concluded that reprecipitation delays the propagation of the reaction front, which reduces the penetration of carbonic acid into the caprock. It has been suggested that precipitation or reprecipitation of the dissolved minerals within the shale matrix may also lead to the so-called "self-sealing of fractures" and to caprock strengthening [139]. The effects of mineral dissolution and reprecipitation on the pore system and permeability of shale have not yet been well described.

Effect of CO2 on Petrophysical Parameters of Shale
This section focuses on the changes in some selected petrophysical parameters of shale after exposure to CO2. It is also important to recall the different states of CO2 while talking about the interactions between CO2 and shale. Pan et al. [140] showed that supercritical CO2 has a significantly greater effect on shale than subcritical CO2. The effects of CO2-brine-shale interactions on porosity, permeability, pore size, and surface area have been reported by several experimental studies [141][142][143][144][145][146][147]. These interactions trigger dissolution, precipitation, and migration of fine clay grains, which can open or close the pore throats [148]. On the one hand, studies have revealed an increase of total porosity and connective porosity upon CO2 exposure [140,142,144]; on the other hand, a decrease of porosity and connective porosity/matrix permeability was also reported [140,142,145,146,[148][149][150]. The increase in pore volume may be related to the decrease in capillary threshold pressure [144]. However, Lahann et al. [121], Okamoto et al. [151], and Gaus et al. [135] reported that there was no, or very little, change in shale's pore structure upon CO2 exposure. Wu et al. [145] explained that this reduction in permeability was dependent on permeability's initial value in shale and the effective stress. Pan et al. [140] investigated a broad range of pore structure parameter changes under sub and supercritical CO2 exposure. It is difficult to highlight a single parameter behind the mechanism that decreases or increases the total porosity, pore diameter/width, and connective porosity under CO2 exposure (especially supercritical CO2). Dissolution and adsorption-induced expansion are the two most important mechanisms able to alter the pore structure during the interactions between subcritical CO2 and shales, but these mechanisms may play different roles in pore structure modification [140]. On the one hand, geochemical reactions dissolve and destroy the internal structures of minerals, leading to nanopore enlargement; on the other hand, adsorption-induced expansion leads to volumetric changes, which may alter the pore structure and ultimately narrow the pores [145,146,152]. Lu et al. [79] reported the swelling of shale samples under CO2 exposure, with an increase and then a drop in stress. This occurs mainly due to a reduction in the surface potential and an increase in the thickness of the surface layer when free CO2 penetrates into pores or cracks. Clay swelling upon CO2 exposure, which may be expected to increase sealing capacity, was also reported by de Jong et al. [153], Geisting et al. [117], and Schaef et al. [154]. However, if the swelling stress exceeds the rock strength, it may destroy the sealing capacity [155]. Pan et al. [140] noticed a difference in the changes of pore structures for marine and terrestrial sediments and concluded that more complicated geochemical reactions may exist between shale and CO2. Moreover, the presence of organic carbon, and its maturity, may affect pore structure modification [156,157]. Since adsorption induced expansion is crucial, it is important to understand the adsorption behaviour of CO2 on shale. Adsorption of CO2 can be varied with pore sizes, temperature and pressure which may influence its induced strain in shale nanopores [107,[158][159][160][161].
The specific surface area is an important parameter that may be linked to the smaller pores of the sample, whereas larger pores are important for diffusion and gas seepage [140,162]. The specific surface area of shale decreases after supercritical CO2 treatment [146]. The increase or decrease of total porosity, connective porosity, number of pores, specific surface area, and permeability depends on the mineralogical composition, the state of the CO2 fluid, and the mechanism acting during CO2 treatments.

The Effect of CO2 on the Mechanical Properties of Shale Rock
As presented previously, the formation of a CO2-plume at the reservoir-caprock interface is usually followed by a pressure increase [14], which will affect the caprock's effective stress and may lead to fault reactivation. CO2-brine-caprock interactions can therefore cause changes in the mechanical properties of rocks, in terms of their strength, stiffness, and fracture thresholds, which can increase the risk of CO2 leaking from the reservoir [163][164][165]. These effects have been experimentally investigated on various shale formations and under different exposure conditions in terms of temperature and pressure [24,70,72,[166][167][168][169][170][171][172][173][174]. The investigated parameters affected by CO2 exposure include volume change (swelling), strength, stiffness, Poisson's ratio, stress thresholds for crack initiation, and crack damage, as well as the brittleness index. The available literature is summarized in Table 1. CO2-water and CO2-brine refer to CO2 in equilibrium with water or brine, respectively. Except for some experiments, the relevant test procedures are globally similar: After exposing the shale samples in a pressure cell to CO2, CO2-water, or CO2-brine under a controlled pressure and temperature for a given saturation time, the samples are brought back to ambient conditions before mechanical and acoustic testing. It seems like the dynamic properties of shales in Al-Ameri et al. [172] were investigated before the samples were brought to an ambient condition. In the experiment of Agofack et al. [70], the sample was subjected to an isotropic pressure of 20 MPa and a pore pressure of 10 MPa. The sample was first saturated with brine, and then mechanical cycles were applied. The mechanical cycle consisted of a decrease in confining pressure from 20 MPa to 15 MPa followed by an increase from 15 MPa to 20 MPa under undrained conditions, with a loading rate of 1 MPa/h. A solution of CO2-saturated brine (CO2-brine), prepared in a separate setup, was flooded at the same pore pressure (10 MPa) for 12 days, and similar mechanical cycles were applied. One should keep in mind that the test procedure can highly affect the experimental results. Indeed, it has been demonstrated that free CO2 can emerge from a solution within a sample if the test pressure is lower than the pressure at which the sample was saturated with CO2-rich brine. The presence of free CO2 affects the compressibility of the pore fluid and the dynamic stiffness of the porous material [175]. These effects were not discussed in the experiments listed here and are not analyzed in this review.
The novelty of the experimental procedure proposed here, as described and used by Agofack et al. [70], lies in the possibility it gives to expose shale rocks to different saturation fluids (brine, dry/wet scCO2, or sbCO2) under a controlled temperature and pressure, without modifying the test exposure conditions prior to mechanical and acoustic measurements of the rock sample. Following that procedure, the sample is placed in a confining chamber under predefined conditions of confining pressure, pore pressure, axial stress, and temperature. Before putting the sample into a rubber sleeve and mounting it into the confining chamber, the lateral surface of the sample is covered by a metal mesh. In addition, two sintered plates are placed on its two sections. The metal mesh and sintered plates are used to enhance the pore fluid distribution across the sections and lateral surface of the sample. A separate new design injection-setup, connected to the pore fluid system of the sample, is used to prepare the CO2 or the mixture of CO2-fluid (water, brine, etc.) at a given temperature and pressure. The injection-setup is then used to flow the prepared CO2 or fluid-rich CO2 into the sample without modifying the experimental conditions of the sample and pore fluid in terms of its temperature and pressure. Before, during, and after the flow of fluid-rich CO2, continuous acoustic measurements of compressive and shear wave velocities, as well as some mechanical loadings, are made. Because the only changed parameter during the experiment is the exposure environment, its effect is better monitored. Some new experimental results using low-frequency measurements [175], which are not presented here, revealed that the Young's modulus of a Draupne shale sample exposed to CO2 increased by around 2%, while its Poisson ratio remained roughly unchanged.

Effect of CO2 on Shale Volume Expansion (swelling) and Brittleness Index
The swelling of shale due to CO2 adsorption has been confirmed in several investigations [79,109,153,176]. The swelling degree depends on the shale's initial smectite content and water saturation. According to de Jong et al. [153], maximum expansion up to 2.46% is reached with an initial interlayer spacing of around 11.1 Å. The largest swelling was observed when the shale samples were exposed to CO2-water [173], probably because under the given conditions of temperature and pressure, the pH of the CO2-water was lower than that of the CO2-brine, making the former more aggressive than the latter [177,178]. Several sorption experiments also showed that adsorptioninduced swelling reduces the mechanical properties of the shale, including its compressive strength [79,173,179,180]. Among the different definitions for the brittleness index, the following was chosen by Lyu et al. [166] in the analysis of their experimental results, which are highlighted here. The brittleness index (BI) is defined as the ratio of reversible strain over the total strain and is given in Equation (4) [181]:

Reversible strain
Total strain The experimental data show that the swelling of shale leads either to an increase or decrease in the brittleness index, depending on whether the shale samples are exposed to only CO2 or to CO2water, respectively [72,166]. According to the results of these authors, dry CO2 seems to induce hardening of the shale material, which is revealed by an increase in the elastic strain or an increase in the brittleness index.

Effect of CO2 on Shale Strength, Stiffness, and Poisson's Ratio
The combined results of the UCS (unconfined compressive strength) and Young's modulus determinations are presented in Figure 5 as a function of exposure time. The different shale samples were exposed to different solutions of CO2, including scCO2, sbCO2, scCO2-brine, and sbCO2-brine (where sc and sb refer to supercritical and subcritical, respectively). These values are normalized by dividing the value at a given exposure time by the value obtained for the non-exposed sample. The normalized numbers, therefore, reflect the percent of reduction. The UCS and Young's modulus can decrease by as much as 60% after 30 days of exposure depending of the type of the exposure fluid.

Exposure to Dry CO2
After exposure to dry sub-and supercritical CO2, the experiments show a gradual decrease in the UCS and static Young's modulus with saturation time [168,171,173]; they also show a decrease in the dynamic Young's modulus [172]. An outcrop anisotropic shale from the Sichuan basin in China was exposed for 10 and 20 days to CO2 and CO2-water at a temperature of 40 °C under a pressure of 9 MPa. When exposed only to CO2, the authors reported a gradual reduction of UCS with exposure time, up to 30% after 20 days [166,169]. The Young's modulus initially showed a reduction up to 36% during the first 10 days of exposure and then started to increase again between 10 and 20 days of exposure [166]. The increase in the Young's modulus after 10 days of exposure to CO2 was confirmed by the same authors in subsequent experiments. A gradual decrease in the Young's modulus was measured within 30 days of exposure to both sub-and supercritical CO2 [169]. The results show that after 10 days of exposure, both the UCS and Young's modulus decreased when CO2-saturation pressure increased from 0 to approximately 12 MPa and then started to increase at a saturation pressure from 12 to 16 MPa [167]. However, when the dry shale samples were exposed to scCO2 for 2 weeks, the Young's modulus and Poisson's ratio were hardly affected. The effect of dry CO2 on the wet samples seems to be higher than that on the dry samples. This is because, on the dry samples, as presented in the previous sections, the effect of water evaporation and that of pH change due to exposure to CO2 are absent, so the impact of CO2 is reduced.

Shale Samples Exposed to CO2-Water or CO2-Brine
All experimental tests on the mechanical properties of shales upon exposure to CO2-saturated fluids showed reduction of their UCS, Young's modulus, and Poisson's ratio with exposure time, as presented in Figure 5 [70,72,168,173]. Reductions up to 66% and 56% were, respectively, measured for the UCS and Young's modulus after 30 days of exposure to the CO2-water mixture [72]. On similar shale samples exposed to gaseous CO2-and scCO2-brine for 7 days, Zhang et al. [168] found reductions up to 32% and 19% for the UCS and Young's modulus, respectively. In the experiment of Agofack et al. [70], a shale sample was subjected to triaxial loading conditions consisting of 20 MPa of confining pressure and 10 MPa of pore pressure. After consolidation, and prior to the flow of CO2brine, two isotropic compression unloading/reloading cycles were applied. After flooding with CO2brine for more than 12 days, similar cycles were performed. With the permeability in the range of nD, drained conditions are very difficult to achieve, and all the cycles were considered as performed under undrained conditions. Agofack et al. [70] also showed a reduction of the Young's modulus and Poisson's ratio, while the maximum deviatoric stress (axial stress minus confining pressure) increased after exposure to CO2-brine. A reduction of the undrained bulk modulus after exposure to CO2-brine was also reported [70]. The cause of this reduction in the experiment of Agofack et al. [70] could be the combined effect of change in the microstructure due to chemical reactions and that of the compressibility of the pore fluid due to free CO2 coming out of solution when the pore pressure decreased during the applied cycles [175]. and CO2-brine. The normalization is done by dividing the value at a given time by the initial value (without CO2). The nomenclature consists of the exposure fluid followed by the references [e.g., scCO2-brine_Zhang et al. [168] refers to the shale sample exposed to supercritical CO2-brine investigated by Zhang et al.].

Stress Thresholds for Crack Initiation and Crack Damage
During the unconfined compressive strength (UCS) tests, the samples were equipped with acoustic emission sensors to record crack initiation and propagation. The signals from the sensors (of ultrasonic frequency) were correlated with the stress state to provide stress thresholds for crack initiation and damage [72,[166][167][168]. The results are presented in Table 2 for different solutions of CO2 and different times of exposure. As presented in the previous section, for a given fluid, the peak stress gradually decreases with exposure time. For a given reaction time, it decreases when pressure is increased between 4 and 12 MPa. For a given exposure fluid under constant temperature and pressure, the stress thresholds for crack initiation and crack damage decrease with the exposure time [72]. For a given exposure time, Yin et al. [167] showed that the stress for crack initiation and crack damage decrease with increasing pressure up to 12 MPa and then increase for pressures above 12 MPa. However, the stress thresholds for crack initiation and damage show a decreasing trend with pressure (between 6 and 12 MPa) when exposed to CO2 or CO2-brine [168]. In other words, as the pH of a CO2-rich fluid decreases with an increase of pressure, the solution becomes more acidic and more aggressive and tends to weaken the shale sample by reducing its peak strength, as well as its thresholds for crack initiation and crack damage.

Tensile Strength and Surface Energy
Brazilian tests (or indirect tensile strength tests) are commonly used to estimate the tensile strength of shale. For a given cylindrical sample of diameter D and thickness h, the following relation is used to calculate the Brazilian splitting tensile strength: where Pmax is the peak load (in Newton) applied to the sample. Equations (6) and (7) show a relationship between the tensile stress and the adsorption induced change in surface tension. These relations can be used to explain the weakening effect resulting from the adsorption of CO2 in shale or clay minerals [164,182]: where σt is the tensile stress at an existing crack tip required to form a new crack surface, E is the Young's modulus, γ is the surface energy, a is the half crack length, and i dμ is the change in the chemical potential of the i th adsorbent component. Equations (6) and (7) refer to Gibbs' and Griffith's relations, respectively. According to Equation (6), the replacement of an existing fluid by another fluid with greater chemical potential ( 0 d μ > ) results in a decrease of the surface energy. Equation (7) shows that this decrease in surface energy causes a reduction in the tensile stress. In addition, the formation of a new crack surface, like during a Brazilian test, contributes to increasing the crack length and again to the reducing the tensile stress. The strength of the material can then be reduced by replacing an existing adsorbate (e.g., brine) with a more reactive fluid with a greater chemical potential (e.g., CO2). Thus, a chemical reaction that leads to a decrease of surface energy in the adsorbate-adsorbent system will reduce the tensile strength of the material. Feng et al. [24] exposed several samples of anisotropic shale from Sichuan in China to scCO2 for a time period of up to 60 days. After exposure and being returned to ambient conditions, samples with different bedding orientations were submitted to Brazilian testing. The results show a gradual decrease of tensile strength with exposure time, and little effect of bedding orientation was found. These results are similar to those found by other authors [171], as presented in Figure 6. The normalized values, also presented in previous figures, were obtained by dividing the value at a given age of exposure over the initial value (without CO2). The results of Ao et al. [171] demonstrate a reduction of around 30% for tensile strength after only 5 days of exposure, while the results of Feng et al. [24] show a reduction of around 8% and 45% after 10 and 60 days, respectively. The shale samples of these two investigations were from the same basin but with different total organic content (TOC). The reduction in tensile strength during the first week of exposure seems to be highly correlated with the TOC. In contrast, the results of Ojala [183] did not show any effect of CO2 on the tensile strength after exposure of Pierre shale to water and CO2-water for one week. Based on the surface energy theory presented in Equations (6) and (7), the results of Figure 6 seem to demonstrate that new cracks were formed within the shale specimens upon their exposure to CO2.  Table 1. The Brazilian splitting modulus, which shows the ability of shale to resist elastic deformation, was also measured; a continuous decrease with exposure time was observed, and little effect with bedding orientation was found [24].

Concluding Remarks
The interactions of shale with CO2 are a complex phenomenon. This complexity arises not only from the diversity of the mineralogical composition of shale rocks but also from their variety of possible interaction mechanisms. A summary of the possible CO2-shale interaction mechanisms discussed in this paper is presented in Figure 7.  Both physical and chemical processes can affect shale's composition, petrophysical parameters, and mechanical behavior at different scales, from nano to macro. The main processes accompanying the exposure of shale to wet CO2 or CO2-saturated brine include the mineral dissolution, precipitation, and reprecipitation of some minerals, as well as volume change in clays resulting from pH and ion concentration changes. Upon exposure to dry CO2, the main processes occurring in shale rocks include shale dehydration, mineral precipitation, dissolution of organic matter (such as kerogen), and the adsorption of CO2. The combination of these processes shows a significant reduction in the mechanical properties of shales in terms of its strength, stiffness, and Poisson's ratio, as well as the stress threshold for crack initiation and crack damage. The largest reductions were measured on wet samples exposed to dry supercritical CO2 or scCO2-saturated brine. The interaction mechanisms underlying the observed changes were, however, not always experimentally addressed. Presently, the effect of dry CO2 on shale desiccation, as well as the effect of supercritical CO2 on kerogen dissolution, and their consequences on porosity, permeability, as well as their mechanical properties, are still not well understood. Multidisciplinary studies are needed to link observed mechanical changes with physicochemical changes in shales. It seems crucial that the interactions of shale with CO2-saturated brine and dry subcritical or supercritical CO2 need to be better described to obtain a complete picture of the potential failure mechanisms. These comprehensive studies should include both batch and flow conditions to mimic equilibrium conditions and far-from-equilibrium conditions. To access the mechanical effects upon exposure to CO2, the available experimental procedures need to be improved. In the majority of investigations, the exposure conditions, in terms of their temperature and pressure, are disturbed before any mechanical measurements, which, by affecting the microstructure of the sample, can add errors to the resulting interpretations. In addition, sample variability should also, as far as possible, be avoided. The experimental procedure should then be as follows: The sample should first be put under in situ conditions in terms of temperature and pressure, and then a fluid similar to its initial pore fluid should be used for re-saturation. After consolidation, the dry CO2 or CO2-rich pore fluid should be flooded without modifying either the sample or the stress and temperature conditions. During these phases, measurements should be made continuously either by using acoustic waves or by applying mechanical cycles on the sample with a stress amplitude less than 1 MPa (to limit mechanical cycle-induced damage).
Author Contributions: In this review paper, M.H.B. described the chemical and petrophysical properties of shale; N.A. described the changes in mechanical properties due to interactions between shale and CO2; K.M.G. described the interaction mechanisms and changes in chemical properties due to shale's exposure to CO2. P.R.C. contributed with his valuable suggestions on the scientific and quality control of the manuscript and also on other parts, including the abstract, introduction, and conclusion. All authors have read and agreed to the published version of the manuscript.