Qualitative and Quantitative Changes of Carbonate Rocks Exposed to SC CO2 (Basque-Cantabrian Basin, Northern Spain)

This study aims at the qualitative and quantitative determination of porosity, mineralogical and textural changes in carbonate rock samples after injection of (i) supercritical CO2-rich brine and (ii) dry supercritical CO2, under similar experimental conditions (P ≈ 75 bar, T ≈ 35 ◦C, 970 h exposure time and no CO2 flow). The studied rocks were sampled in the western Basque-Cantabrian Basin, North Spain, and consist of vuggy carbonates (“Carniolas”) of the Puerto de la Palombera formation (Hettangian). Mineralogical and pore space characterization is completed using optical microscopy, scanning electron microscopy and optical image analysis. In addition, X-ray fluorescence analyses are performed to refine the mineralogical information and to obtain whole rock geochemical data and the brine composition is analysed before and after the experiment. Mineralogical and chemical results indicate that the carbonate rocks exposed to supercritical CO2 in dry conditions do not suffer significant changes. However, the injection of supercritical CO2-rich brine induces chemical and physical changes in the rock due to the high reactivity of calcite at the low pH conditions produced by the acidified brine. Numerical modelling validates the experimental observations. These results can be used to characterize the behaviour of carbonate rocks under conditions similar to the vicinity of a CO2 injection well. The results should be considered only at the scale of the studied samples and not at reservoir scale.


Introduction and Objectives
Reduction of greenhouse gas concentration in the atmosphere has been in the focus of scientific research over the last decades. One of the possible approaches is to capture CO 2 from large industrial sources and to storage it in deep geological formations [1][2][3][4]. During this process, the CO 2 is injected into a formation of high porosity and permeability (reservoir), which is overlaid by an impermeable formation (seal).
Sedimentary rocks-both of siliciclastic and carbonate composition-are especially suitable reservoirs due to their textural and mineralogical characteristics. The high porosity and permeability of many sedimentary formations are decisive factors to consider for the feasibility of CO 2 injection on a large scale [5]. The CO 2 injection is intended in a dense phase or in supercritical conditions (SC CO 2 : P > 73.8 bar, T > 31.7 • C), which results in a significant volume reduction of the gas [1,6]. These conditions occur naturally in reservoirs under 800 m depth.
Injecting CO 2 into geological formations provokes a series of physical and chemical changes, which can significantly modify the reservoir properties [7]. Geochemical changes arise from the

Geological Setting
The main criterion for the site selection of this study was to sample sedimentary formations with real potential to be used as CO2 reservoirs (formations with suitable depth, permeability and with a demonstrated seal formation). The Basque-Cantabrian basin (northern Spain) (Figure 1) was selected according to the results of the research conducted by the Geological Survey of Spain (IGME) in CO2 geological storage [29] and of previous studies concerning hydrocarbon exploration in the same areas [30]. The selection was further refined to the Hontomín structure area [31,32], which has been widely studied and proved to have potentiality for geological storage due to the presence of early structural traps.
The study focuses on the carbonate deposits of the Jurassic (Lias and Dogger) marine sequence of the western part of the Basque-Cantabrian basin [33][34][35][36]; more precisely, on the evaporitic-stromatolithic facies belts of the shallow carbonate ramp, which have suitable porosity and permeability for CO2 storage [37,38]. In this area, the following sequence of facies belts has been described: (i) sabkha facies characterized by dolomites, marls and interbedded anhydrites, forming collapse breccias ("carniolas") on the surface due to the dissolution of the evaporitic components (Puerto de Palombera Fm., Hettnanigian); (ii) peritidal facies formed by stromatolites at the base and oolitic grainstones at the top (Villanueva de Puerta Fm, Lower Sinemurian); (iii) open ramp facies characterized by cycles of mudstone to grainstone sequences (Sopeña Fm., Lower Sinemurian); and finally, (iv) fluvial to shallow marine mixed carbonate and siliciclastic deposits (Río Polla Fm., late Lower Sinemurian) [35]. The materials of the shallow carbonate ramp are overlaid unconformably by alternations of marls, marly limestones and shales of a deeper (100-400 m) hemipelagic ramp [36]. The studied rocks were sampled in the south-western part of the Basque-Cantabrian basin and consisted of vuggy carbonates ("Carniolas") of the Puerto de la Palombera formation.

Methodology
The overall methodology to study the mineralogical and geochemical changes of carbonate rocks when exposed to SC CO2 rich brine consisted of the following steps:


Selection of a study area with real potential for CO2 geological storage ( Figure 1).  Identification of reservoir rocks in this area and their sampling for laboratory experimentation ( Figure 1).  Preparation of rock samples for laboratory experiments providing sample pairs in order to conduct comparative studies before and after SC CO2 exposure. The sample pairs are prepared to have one matching surface as close to each other in the original block as possible ( Figure 2).  Mineralogical and chemical characterization of rock samples and definition of brine composition before the experiment.

Methodology
The overall methodology to study the mineralogical and geochemical changes of carbonate rocks when exposed to SC CO 2 rich brine consisted of the following steps:

•
Selection of a study area with real potential for CO 2 geological storage ( Figure 1). • Identification of reservoir rocks in this area and their sampling for laboratory experimentation ( Figure 1). • Preparation of rock samples for laboratory experiments providing sample pairs in order to conduct comparative studies before and after SC CO 2 exposure. The sample pairs are prepared to have one matching surface as close to each other in the original block as possible ( Figure 2

Experimental Setup (Sample Preparation and SC CO2 Injection)
Sample preparation plays a crucial role in experimental studies especially when the analytical procedure is based on a comparative study of samples. In order to minimize the effect of rock heterogeneity, the mineralogical (OpM-SEM) and geochemical (XRF) characterization of the samples before and after CO2 exposure was conducted on neighbouring areas of various blocks of 27 cm 3 . Although the thin sections and the SEM specimens of before-after samples obviously could not be prepared of exactly the same sample area, they were taken from surfaces as close to each other as possible (a few millimetres of distance). For the geochemical comparative analyses neighbouring blocks were used. The sample blocks selected for CO2 exposure were saturated with brine in cycles of 72 h under atmospheric conditions.
The experiment was conducted in a static Autoclave chamber (without CO2 flow during the test) at 7.5 MPa pressure, 35 °C temperature, during 970 h. The configuration of the experimental setup (chamber, pumps, CO2 cylinders, connections, etc.) was based on other systems described by [15,20,26,28]. In general terms, two cylinders of commercial CO2 of 99.8% purity were connected via 5 mm stainless steel tubes to (i) the chamber of 3 dm 3 where the samples were placed and (ii) to a piston pump also connected to the chamber. The first cylinder, directly connected to the chamber, makes it possible to fill the chamber with CO2 at atmospheric temperature and 4.5 MPa pressure. The second cylinder allows the user to pressurize the chamber applying a controlled gas flow until the desired pressure value (i.e., 7.5 MPa). The inside material of the chamber is Hastelloy C276 alloy.

Experimental Setup (Sample Preparation and SC CO 2 Injection)
Sample preparation plays a crucial role in experimental studies especially when the analytical procedure is based on a comparative study of samples. In order to minimize the effect of rock heterogeneity, the mineralogical (OpM-SEM) and geochemical (XRF) characterization of the samples before and after CO 2 exposure was conducted on neighbouring areas of various blocks of 27 cm 3 . Although the thin sections and the SEM specimens of before-after samples obviously could not be prepared of exactly the same sample area, they were taken from surfaces as close to each other as possible (a few millimetres of distance). For the geochemical comparative analyses neighbouring blocks were used. The sample blocks selected for CO 2 exposure were saturated with brine in cycles of 72 h under atmospheric conditions.
The experiment was conducted in a static Autoclave chamber (without CO 2 flow during the test) at 7.5 MPa pressure, 35 • C temperature, during 970 h. The configuration of the experimental setup (chamber, pumps, CO 2 cylinders, connections, etc.) was based on other systems described by [15,20,26,28]. In general terms, two cylinders of commercial CO 2 of 99.8% purity were connected via 5 mm stainless steel tubes to (i) the chamber of 3 dm 3 where the samples were placed and (ii) to a piston pump also connected to the chamber. The first cylinder, directly connected to the chamber, makes it possible to fill the chamber with CO 2 at atmospheric temperature and 4.5 MPa pressure. The second cylinder allows the user to pressurize the chamber applying a controlled gas flow until the desired pressure value (i.e., 7.5 MPa). The inside material of the chamber is Hastelloy C276 alloy.
The temperature inside the chamber (35 • C during the experiment) is controlled by a temperature regulator. All of the elements and sensors (of pressure and temperature) of the system are connected to a central computer which controls and regulates the system setup by the Hel 5.1 software (HEL Ltd., Borehamwood, UK).
Six previously saturated rock samples were placed into the chamber and covered with 0.3 dm 3 of brine. For the dry experiment the samples were not saturated and no brine was introduced into the system. In both wet and dry configurations, the test consisted of three distinct phases: (i) the fill-in phase, filling the chamber with CO 2 until reaching supercritical conditions (3 h); (ii) the experimental phase, during which the supercritical pressure and temperature values are maintained steady; and (iii) the evacuation phase, during which the system is de-pressurized and the fluid and solid samples are collected (3 h). All the experimental runs for this study were carried out in the Laboratories of the IGME in Tres Cantos, Madrid, Spain.

Applied Techniques and Methods
A detailed petrographic description was the first step of the study using optical microscopy (OpM) (Leica DM 6000, 20×, 10×, Wetzlar, Germany). This study was repeated after the SC CO 2 exposure. In order to obtain more detailed qualitative data about the rock petrography, Scanning Electron Microscopy was also applied (JEOL 6100 SEM, using W-Filament, acceleration voltage of 20 kV and Inca Energy-200 software (Inca Energy, Abingdon, UK)). By means of optical image analysis (OIA) the main parameters of optical porosity were quantified using ImageJ software on OpM images acquired by a Charge-Coupled Device (CCD) camera (ProgRes 5.0, Jena, Germany). The chemical composition of the brine was determined by the combination of ion chromatography, inductively coupled plasma optical emission spectrometry (ICP-OES), pH and conductivity measurements, both before and after the experiment. Similarly, the whole rock chemical composition was also measured twice (before and after SC CO 2 exposure) by XRF. These studies were developed in the facilities of the IGME and the University of Oviedo (UNIOVI). The mineral composition of the initial rock sample used in this study was determined from the XRF analyses performed at the IGME laboratory. To do so, we used Excel solver in order to determine the optimal mineral composition corresponding to the XRF analyses and the rock description. This mineral composition as well as the injected water solution were used as input for the mass balance calculation and the numerical modelling.
The geochemical modelling was performed using the CrunchFlow code [39]. The CrunchFlow code solves numerically the mass balance of solutes solving the reactive transport equation. This code has been validated by several studies focused on CO 2 storage [40][41][42][43]. The total reaction rate R i is given by (1): and ν im is the number of the moles of i in mineral m. The reaction rate laws used in the calculations are of the form (2): where A m is the mineral surface area in m 2 m −3 rock, k m,T is the kinetic rate in mol m −2 s −1 , a n H+ H+ is the term describing the effect of pH on the rate, a ni i is the term describing a catalytic/inhibitory effect on the rate by other species, ∆G r is the Gibbs free energy of the reaction (J mol −1 ), IAP is the ionic activity product of the solution with respect to the mineral, K eq is the equilibrium constant for that mineral reaction (ionic activity product at equilibrium) and m 1 and m 2 exponents allow for nonlinear dependencies on the affinity term and are normally taken from experimental studies. The summation term indicates that several parallel rate laws may be used to describe the dependence of the rate on pH or on other species. The kinetic rate at temperature T (K) is calculated from (3): where k m,25 is the kinetic rate at 25 • C, E a is the apparent activation energy of the overall reaction (J mol −1 ) and R is the gas constant (J mol −1 K −1 ). Change in mineral surface area A m (m 2 mineral m −3 rock) is calculated assuming that area is proportional to the volume fraction of the mineral to the power 2/3. The numerical model developed to reproduce the experimental conditions ("batch" simulation) considered (a) the mineralogical composition of the rock according to the measured values by XRF and (b) the rate laws for the reacting minerals showed in the literature [44][45][46][47].
The best fit of the model to the experimental data (aqueous Ca, S, Mg, K, Si, Al concentrations) was performed by adjusting the values of the effective reactive surface areas of each mineral, which is a parameter that is highly difficult to evaluate and was treated here as a simple dummy fitting parameter to account for unmodelled mass transfer processes. Regarding the secondary minerals, the value of the initial surface area was set to 1000 m 2 /m 3 for initiating the precipitation.
Preliminary simulations were done for determining the secondary minerals that are effectively precipitating among those given in the thermodynamic data base. The water inlet in the model is the same as the one used in the laboratory experiment. The calculated initial pH is 3.32. All the equilibrium constants (log K at 25, 40 and 60 • C) and stoichiometric coefficients were taken from the EQ3/6 database [48]. Activity coefficients were calculated using the extended Debye-Hückel formulation (b-dot model) with parameters listed in the database. The equilibrium constants for the mineral reactions were also taken from the EQ3/6 database.

Petrographic and Chemical Characterization of Carbonate Rocks before SC CO 2 Injection
The Puerto de Palombera Formation mainly consists of a heterogeneous, recrystallized carbonate breccia that presents a high percentage of macroporosity (≈30%) in surface outcrops due to the dissolution of evaporate minerals. The partial recrystallization resulted in a matrix of highly varying grain size from micrite up to crystals of 200 µm. The angular-subangular clasts are composed of sparite and could be intra-, extra-or bioclasts. Their size varies between 0.1 mm and >2 mm (Figure 3a). The high calcite/dolomite ratio and the textural features suggest a dolomitization, followed by a posterior de-dolomitization process (Figure 3b). Calcite filled veins and pores are frequently observed and open fractures and intercrystalline porosity affect both fragments and matrix (Figure 3c). Geode type open pores are also present (Figure 3d). The optical porosity measured by optical image analysis (OIA) on thin sections is ≈4.1 ± 0.15% and the average pore size is (≈20 um).
The major element composition of the samples before CO 2 exposure was determined by XRF (Table 1 Rock before CO 2 column). Table 1 shows the values in weight % and their uncertainty range.
The optimal mineral composition corresponding to the XRF analyses and the rock description was obtained applying numerical correction. The initial mineral composition of the studied rock was 92. 8    On the thin section scale, no changes (either dissolution or precipitation) could be detected by OpM and OIA on the carbonate samples after their exposure to dry CO2. The optical properties of the rock forming minerals did not present any kind of alteration, neither was the value of the optical  On the thin section scale, no changes (either dissolution or precipitation) could be detected by OpM and OIA on the carbonate samples after their exposure to dry CO 2 . The optical properties of the rock forming minerals did not present any kind of alteration, neither was the value of the optical porosity modified by the experiment. We can confirm thus that, at least at the scale of the microscopic observations, the mineralogical and textural configuration of the rock remained intact and none of the possible processes (e.g., pore filling by precipitations, increased surface roughness by dissolution, recrystallization, etc.) took place. Observations carried out with SEM also showed a lack of evidence of dissolution or precipitation of the main mineral components. The Cal/Dol ratio remained the same as in the original samples and the qualitative assessment of the calcite and dolomite grain morphology did not show any evident shape changes provoked by the CO 2 exposure. The analysis of the chemical composition of samples after dry CO 2 exposure (Table 1 "Rock after dry CO 2 " column) provided values within the uncertainty ranges of the original values, which again confirms the non-existence of reactions between the rock forming minerals and the dry SC CO 2 .

SC CO 2 Rich-Brine
The composition of the brine was analysed using a combination of techniques before the experiment ( Table 2 "Pure Brine" column). The brine that was used in the experiment was recovered immediately after the end of the test for a comparative chemical analysis. The results are given in Table 2  The optical observations carried out by OpM and SEM (Figure 4) demonstrate obvious changes in the calcite mineral grains as compared to the natural sample. In particular, dissolution textures can be observed including (i) the apparition of secondary pores on the surface of calcite minerals and (ii) a relatively rough surface after CO 2 -brine-rock interaction (Figure 4b). Quantification of the optical pore parameters by OIA revealed an increase in number of pores (12.26%) and in total pore area (4.12% to 4.44%) due to the formation of new pores and the increase in size of the existing ones. Pore morphology was also affected as shown by the increased aspect ratio and the decrease in the roundness parameter (Table 3). The XRF analysis however did not reveal significant changes in the whole rock composition after the experiment (Table 1 "Rock after CO 2 and brine" column). Table 2. Chemical analysis of pure brine and brine taken from the reaction chambers. Uncertainty (10%) given by IGME Laboratory.  Table 3. Number of pores, optical porosity, roundness (Ro) and aspect ratio (As) measured by OIA of carbonate thin sections before and after SC CO 2 exposure. (∆: relative variation; * absolute variation; ** uncertainly 1.26% given by Oviedo-IGME OIA Laboratory).

Mass Balance Calculation and Numerical Modelling
The mass balance calculation indicates dissolution of calcite, dolomite, quartz and pyrite and precipitation of illite and gypsum. The total amount of dissolved and precipitated minerals as well as the final mineral composition is presented in Table 4.
The mass balance has been done assuming that changes of Mg concentration in the fluid are due to dolomite dissolution only and no Mg-rich mineral precipitated. The mass of dissolved pyrite is

Mass Balance Calculation and Numerical Modelling
The mass balance calculation indicates dissolution of calcite, dolomite, quartz and pyrite and precipitation of illite and gypsum. The total amount of dissolved and precipitated minerals as well as the final mineral composition is presented in Table 4.
The mass balance has been done assuming that changes of Mg concentration in the fluid are due to dolomite dissolution only and no Mg-rich mineral precipitated. The mass of dissolved pyrite is calculated using the Fe concentration and the gypsum amount is evaluated by sulphate concentration changes after the pyrite calculation. Quartz dissolution is calculated using the Si concentration changes while the amount of illite is evaluated using the amount of K concentration. No precipitations of other clay minerals such as kaolinite are considered here, even if in other recent studies kaolinite precipitation has been considered the most favourable secondary alteration mineral [49][50][51]. Here, we stipulate that without alumino-silicate in the initial rock sample such as albite or microcline, kaolinite precipitation is not expected. Moreover, illite is initially identified in the rock sample composition which can accelerate the nucleation step.
Regarding the results of the numerical modelling, we can observe in Table 4 that the simulated results are similar to those obtained by the mass balance calculation. The main reaction is calcite dissolution even if the saturation index for calcite was not as low as for dolomite ( Figure 5). This difference is explained by the calcite dissolution rate (kinetics and reactive surface area). Illite precipitation is also predicted by the model as calculated from the mass balance. Actually, the saturation index for illite remains quite high at the end of the simulation indicated that equilibrium is not yet reached and more illite precipitation may occur. In the simulation results, we can observe that calcite and gypsum have reached equilibrium at the end of the experiment, while dolomite is still under-saturated with respect to the fluid. The secondary minerals that were finally used in the calculations are given in Table 4 and are in agreement with the ones supposed for the mass balance calculation. The kinetics data presented in Table 4 and used in this study are in agreement with other studies performed in similar temperature and pressure conditions [27,43,[50][51][52]. Table 4. Mineral composition before and after the batch experiment obtained from mass balance calculation and numerical modelling. The geochemical (thermodynamical and kinetics) parameters used in the CrunchFlow simulation are also indicated as well as the references. calculated using the Fe concentration and the gypsum amount is evaluated by sulphate concentration changes after the pyrite calculation. Quartz dissolution is calculated using the Si concentration changes while the amount of illite is evaluated using the amount of K concentration. No precipitations of other clay minerals such as kaolinite are considered here, even if in other recent studies kaolinite precipitation has been considered the most favourable secondary alteration mineral [49][50][51]. Here, we stipulate that without alumino-silicate in the initial rock sample such as albite or microcline, kaolinite precipitation is not expected. Moreover, illite is initially identified in the rock sample composition which can accelerate the nucleation step.

Mineral and Chemical Composition
Regarding the results of the numerical modelling, we can observe in Table 4 that the simulated results are similar to those obtained by the mass balance calculation. The main reaction is calcite dissolution even if the saturation index for calcite was not as low as for dolomite ( Figure 5). This difference is explained by the calcite dissolution rate (kinetics and reactive surface area). Illite precipitation is also predicted by the model as calculated from the mass balance. Actually, the saturation index for illite remains quite high at the end of the simulation indicated that equilibrium is not yet reached and more illite precipitation may occur. In the simulation results, we can observe that calcite and gypsum have reached equilibrium at the end of the experiment, while dolomite is still under-saturated with respect to the fluid. The secondary minerals that were finally used in the calculations are given in Table 4 and are in agreement with the ones supposed for the mass balance calculation. The kinetics data presented in Table 4 and used in this study are in agreement with other studies performed in similar temperature and pressure conditions [27,43,[50][51][52]. Table 4. Mineral composition before and after the batch experiment obtained from mass balance calculation and numerical modelling. The geochemical (thermodynamical and kinetics) parameters used in the CrunchFlow simulation are also indicated as well as the references.

Discussion
SC CO2 injection into a carbonate reservoir rock was studied experimentally in two distinct situations. The dry experiment did not produce any changes that could be detected by the applied

Discussion
SC CO 2 injection into a carbonate reservoir rock was studied experimentally in two distinct situations. The dry experiment did not produce any changes that could be detected by the applied methodology. This is in agreement with the preliminary expectations based on theoretical models and results of similar experiments [3,20,24]. The absence of available pore water precludes the development of a low pH environment and the consequent carbonate dissolution; while other processes possibly provoked by dry CO 2 injection, such as drying of clay minerals or particle movement, were not expected in this case due to the high carbonate content of the rock and the static nature of the experimental setup. Dry or almost dry conditions are possible to evolve in near-well situations where the injected CO 2 displaces most of the pore water and the high CO 2 -brine ratio results in the dissolution of the remaining (e.g., capillary) water [7]. Under these conditions the chemical reactions are not really relevant and any changes induced in the pore configuration will be due to other processes.
On the other hand, with increasing distance from the injection point, the amount of pore water increases [7] and the chemical reactivity of the system becomes a dominant factor. Our experiment modelling water saturated conditions produced obvious chemical and physical changes in the reservoir rock.
Therefore, calcite dissolution is usually expected under these acidic conditions. Nevertheless, the analysis of whole rock composition by XRF (Table 1) did not indicate any dissolution or precipitation reactions in this case. This is probably due to the sample scale of the analysis. The dissolution effects are expected to be rather superficial while for the XRF measurement a 27 cm 3 block was used. The total amount of dissolved minerals is 271.5 mm 3 (Table 4), which is 0.25% of the initial volume of rock (4 blocks of 27 cm 3 ). This quantity is very low and validates the results of the XRF acquired on the full rock blocks. Moreover, the high value of CaO in the rock composition also entails a high uncertainty range (~3.5%), which makes it difficult to evince small changes in the calcite composition by this technique. Nevertheless, the dissolution reactions are probably localized at the block surfaces.
On the other hand, the changes of water composition clearly indicate dissolution reactions (see Table 2). The obtained data demonstrates the chemical instability of calcite and the relative stability of dolomite. Comparative studies on dissolution rates of different carbonates under similar conditions has shown higher dissolution rates for calcite and higher quantities of dissolved Ca 2+ than Mg 2+ in dolomites [55,56], which is in agreement with our findings.
Additionally, physical changes were observed in the pore network configuration by microscopic techniques including an increase in porosity and changes in pore morphology (Figure 4). Dissolution surfaces and newly formed pores were observed on calcite grains but not on dolomite, although the scarcity of dolomite crystals in the rock makes it difficult to carry out thorough observations. The quantification of pore parameters (optical porosity, aspect, roundness) by means of OIA did not reveal highly significant changes. It is important to point out, however, that while the qualitative observations were conducted both at thin section (OpM) and SEM scales, the quantification was only carried out on thin sections. Evidences of dissolution processes were mostly found by SEM observations so the measurements carried out by OpM clearly provide lower values of porosity and pore morphology changes than the reality. The obtained values should be considered keeping in mind this deviation.
The described changes could be illustrated by the following simplified conceptual model (Figure 6a-d). The CO 2 -rich brine reacts with the sample producing chemical and physical changes on the external surface of the carbonate blocks. These effects are a result of partial carbonate dissolution affecting calcite grains and carbonate cement. Appl. Sci. 2017, 7, 1124 12 of 17 on the external surface of the carbonate blocks. These effects are a result of partial carbonate dissolution affecting calcite grains and carbonate cement. In order to validate the obtained results and the conceptual model, mass balance calculation and numerical modelling was applied. Using the initial mineral composition of the rock sample determined by the XRF measurement and the water chemistry, we evaluated the quantity of dissolved and precipitated minerals and identified the dissolution and precipitation reactions that occurred during the 970 h of CO2 rich water exposure. The numerical results are in agreement with the XRF data and water chemistry analysis. Both indicate calcite and dolomite dissolution coupled with secondary gypsum and illite precipitation.
The described study is limited to two specific situations within the extremely complex system of CO2 storage in geological structures. The time and space scales, the variability of conditions and the number of components of a laboratory experiment are very far from the real-life conditions. The discussed reactions may trigger a whole set of additional processes including the precipitation of new mineral phases due to the increased ion concentration of the brine [25,57], the adjustments in brine mobility due to changes in the capillary network [21] or the alteration of mechanical properties of the rock provoked by the changes in the pore network [58,59], to name just a few. Nevertheless, our results provide some insight into the physico-chemical processes that may play a role in a CO2 reservoir. It is important to take into account that the results obtained in this research must be In order to validate the obtained results and the conceptual model, mass balance calculation and numerical modelling was applied. Using the initial mineral composition of the rock sample determined by the XRF measurement and the water chemistry, we evaluated the quantity of dissolved and precipitated minerals and identified the dissolution and precipitation reactions that occurred during the 970 h of CO 2 rich water exposure. The numerical results are in agreement with the XRF data and water chemistry analysis. Both indicate calcite and dolomite dissolution coupled with secondary gypsum and illite precipitation.
The described study is limited to two specific situations within the extremely complex system of CO 2 storage in geological structures. The time and space scales, the variability of conditions and the number of components of a laboratory experiment are very far from the real-life conditions. The discussed reactions may trigger a whole set of additional processes including the precipitation of new mineral phases due to the increased ion concentration of the brine [25,57], the adjustments in brine mobility due to changes in the capillary network [21] or the alteration of mechanical properties of the rock provoked by the changes in the pore network [58,59], to name just a few. Nevertheless, our results provide some insight into the physico-chemical processes that may play a role in a CO 2 reservoir. It is important to take into account that the results obtained in this research must be interpreted within the context and scale of the described work (i.e., laboratory test). The study of a significantly larger number of samples together with the modelling data following the presented approach would make it possible to gain a clearer view of the precipitation and dissolution processes at a larger (i.e., reservoir) scale. This would also provide important information about the potential of the study area for geological CO 2 storage. Flow-through experiments might also be an alternative option to characterize the chemical reaction during CO 2 injection taking into account the transport effect. Finally, numerical modelling at reservoir scale integrating these experimental results could predict the potential CO 2 sequestration in this reservoir.
The real importance of the encountered mineralogical and porosity changes is in their influence on the functionality and integrity of a storage site. Slight textural changes in the reservoir or seal rock might lead to significant modifications in the storage capacity, injection rate, geomechanical resistance, confining capacity, etc. of the whole system. To study these overall processes on more realistic time and space scales, numerical modelling has been successfully applied [4,7,[12][13][14][15] adjusting the models with experimental data such as those provided by this work.
Migration of fine particles induced by either chemical or physical forces has also been described in various fields of geological reservoir use. In general, the processes and the overall effect of particle migration on the hydrodynamic properties of the rock vary depending on characteristics of the rock and the circumstances. Some authors [60][61][62][63][64] have studied the importance of fine particle migration in CO 2 storage and during brine disposal demonstrating its significance on flow pathways and engineered water injection. For more detailed studies, migration of fine particles should be considered.

Conclusions
Experimental CO 2 injection into carbonates with real CO 2 storage potential was used to simulate possible interactions in the rock-gas system in dry and in water saturated conditions. Both of these conditions can evolve in the vicinity of the injection well during the first phases of injection. The experiments carried out in a hyperbaric chamber provide valuable insight about the processes although they were limited to specific cases. The tests presented here (approximately 1000 h) demonstrate that in the absence of available pore water chemical reactions do not take place in the SC CO 2 -carbonate system. On the other hand, under water saturated conditions, significant mineralogical and physical changes were observed in the rock at thin section scale (OpM, OIA and SEM) and in the brine, even though they were not confirmed by whole rock chemical analysis. In particular, the exposure of carbonate rocks to SC CO 2 rich brine under static conditions resulted in: (i) The dissolution of CO 2 in the brine, which created an acidified environment (HCO 3− increase in brine: 1400%; pH decrease). (ii) Calcite dissolution in the low pH brine (Ca 2+ increase in brine: 290%), producing dissolution phenomena observable by microscopic techniques (OpM and SEM). (iii) Increased porosity (absolute porosity variation: 0.32%) and modified pore morphology due to the calcite dissolution (Ro decrease: 2.40%; As increase: 2.26%; and increase of pore number: 12.26%). (iv) Dolomite crystals also present in the rock were found to be less reactive than the calcite.
(v) Illite and gypsum precipitation was evaluated by both mass balance calculation from the water chemistry and numerical modelling.
The laboratory experiment and numerical modelling performed in this study indicate that calcite dissolution is the main reaction. One of the most important results is that equilibrium between the fluid and calcite is reached in around 50 h. This dissolution reaction induces a slight increase of