CO 2 -Fluid-Rock Interactions and the Coupled Geomechanical Response during CCUS Processes in Unconventional Reservoirs

The di ﬃ culty of deploying remaining oil from unconventional reservoirs and the increasing CO 2 emissions has prompted researchers to delve into carbon emissions through Carbon Capture, Utilization, and Storage (CCUS) technologies. Under the con ﬁ nement of nanopore in unconventional formation, CO 2 and hydrocarbon molecules show di ﬀ erent density distribution from in the bulk phase, which leads to a unique phase state and interface behavior that a ﬀ ects ﬂ uid migration. At the same time, mineral reactions, asphaltene deposition, and CO 2 pressurization will cause the change of porous media geometry, which will a ﬀ ect the multiphase ﬂ ow. This review highlights the physical and chemical e ﬀ ects of CO 2 injection into unconventional reservoirs containing a large number of micro-nanopores. The interactions between CO 2 and in situ ﬂ uids and the resulting unique ﬂ uid phase behavior, gas-liquid equilibrium calculation, CO 2 adsorption/desorption, interfacial tension, and minimum miscible pressure (MMP) are reviewed. The pore structure changes and stress distribution caused by the interactions between CO 2 , in situ ﬂ uids, and rock surface are discussed. The experimental and theoretical approaches of these ﬂ uid- ﬂ uid and ﬂ uid-solid reactions are summarized. Besides, de ﬁ ciencies in the application and safety assessment of CCUS in unconventional reservoirs are described, which will help improve the design and operation of CCUS.


Introduction
At present, conventional oil and gas production is decreasing over the years, and the unconventional oil and gas resources, which with rich reserves, have attracted more and more attention. The past decade has seen breakthroughs in North American shale oil development with horizontal wells' maturity and large-scale hydraulic fracturing. The successful development of unconventional resources has transformed the world's energy landscape. Despite advanced horizontal well drilling technology and volume fracturing technology, shale oil's primary recovery is extremely low (~5%-10). The low recovery is mainly because the adsorbed and dissolved crude oil in the shale reservoir cannot be effectively utilized in the primary development [1]. Simultaneously, clay expands when surfactant and water are injected, and the resulting decrease in permeability may make it challenging to increase reservoir pressure [2]. Researchers experimentally compared the effects of CO 2 Huff-n-Puff and water injection on the enhanced recovery of shale oil. The results show that under different shale properties, CO 2 Huff-n-Puff can increase the recovery rate by~50% (from 33% to 85%), and the overall effect is far better than that of surfactant and water injection [3]. Compared with the injection of other miscible gases such as N 2 and CH 4 , CO 2 has lower minimum miscibility pressure (MMP), resulting in more extensive oil expansion, more viscosity reduction, and mobility improvement. Adsorbed oil can also be effectively used through extraction [4]. CO 2 injection has become one of the potential means of developing unconventional resources efficiently.
Simultaneously, the coupling of scientific and engineering issues with social and political discussions concerns the challenges posed by the harmonious coexistence of energy, climate, and environment. Another research topic that arises is reducing carbon emissions through Carbon Capture, Utilization, and Storage (CCUS) technologies. Over half of the top 10 largest CO 2 storage projects counted in 2019 transported the captured CO 2 by pipeline to nearby oilfields to enhance gas/oil recovery. The Century Plant and Shute Greek Gas Processing Plant projects, which ranked first and second worldwide, have unconventional reservoirs as their primary storage sites. The CO 2 from Century Plant is for use in EOR projects in the Permian Basin. It is expected to use at least 3.5 trillion cubic feet of CO 2 in Avalon shale, Bone Spring, Wolfbone/Wolfcamp, Cline shale, Wolfberry, and Delaware to develop approximately 500 million barrels of reserves from currently owned assets at an attractive cost. The Rangely Weber unit in Salt Creek field, one of the storage sites for Shute Greek Gas Processing Plant, is the world's largest extralow permeability sandstone reservoir developed by CO 2 -WAG miscible flooding [5]. China has included CCUS in its latest five-year economic development plan, is building large-scale carbon storage facilities, and provides financial support for CCUS pilot projects. Notable examples of the developments over the past decade include the CCUS facility in PetroChina Jilin Oil Field, Changqing Oil Field, and Sinopec Zhongyuan [6].
Successful application of CO 2 enhanced recovery (EOR) technology and its storage in unconventional reservoirs requires a comprehensive understanding of CO 2 -fluid-rock interactions, incredibly nanoscale fluid-rock interactions that play an essential role in unconventional formations. Possible interactions in the reservoir within injected CO 2 have been central to the system's overall evolution in dissolution, precipitation, secondary mineral formation, and deposition. This paper mainly reviews the in situ fluid between CO 2 and reservoir brine, and crude oil and the physical and chemical reactions between CO 2 and reservoir minerals. The pore size changes, connectivity, and reservoir stress distribution caused by fluid-fluid and fluid-solid reactions are also summarized. Finally, the future research directions and challenges for the research and application of CO 2 -CCUS are discussed.

CO 2 -Oil Interactions and Phase Behaviors
2.1. Adsorption and Desorption. As early as the beginning of the last century, it had been established that fluids could adsorb on solid surfaces [7], and this phenomenon was also confirmed in subsequent studies [8,9]. In conventional reservoirs, pores are measured in millimeters. In such porous media with relatively large pore size, fluid adsorption is usually negligible. However, when the pore size in the shale is nanoscale (1-100 nm), the pore space occupied by the adsorbed fluid can no longer be ignored due to the increase of the specific surface area pore. Although there have been many studies on oil and gas adsorption, the performance of the mixture in unconventional reservoirs is still a challenging research topic due to the complex conditions caused by multicomponent fluids, high temperature, high pressure, and heterogeneous pore distributions in reservoirs.
In terms of experiments, microfluidic technology development makes it possible to measure the PVT parameters of multiphase fluids quickly. Molla and Mostowfi [10] completed the PVT test of reservoir fluids in a wide range of <86 MPa and <150°C by making a curved narrow singlecapillary channel model. Rapoport et al. [11] carried out a two-phase flow experiment using a microfluidic model and pointed out that the phase interface can promote two-phase fluid's simultaneous motion. Karadimitriou et al. [12] observed two-phase distribution characteristics in quasistatic displacement and imbibition and measured the phase saturation and interface area. Another factor to consider is the strong interaction between CO 2 and organic matter. The injected CO 2 hydrocarbon fluids are competitive for adsorption in porous media. Chareonsuppanimit et al. [13] carried out CH 4 , CO 2 , and N 2 adsorption experiments on shale and coal and found that the adsorption amount of these gasses on shale was lower than that of coal with the same maturity. Zhang et al. [14] conducted competitive adsorption experiments on shale sing CO 2 -CH 4 binary gas and showed that the separation coefficient would decrease with CO 2 concentration. Bergaoui et al. [15] confirmed that the adsorption n-alkane series, C 1 -C 7 , on activated carbon was much different because of the distinct molecules' sizes. Although the above works are of guiding significance for the competitive experiments, they are mostly focusing on the interactions between CO 2 and light components with short molecular chain length.
Some studies on the simulation of oil/gas adsorption in unconventional reservoirs can be seen in recent years. Molecular simulation studies have predicted the density distribution of methane in slit nanopores [16]. The results showed that methane is adsorbed on the graphite pore wall with a size of 1.14-3.93 nm at the reservoir temperature (353 K) and pressure (21 MPa). Studies of the behavior of binary mixtures (C 1 /C 2 and C 1 /C 3 ) using the Grand Canonical Monte Carlo (GCMC) method have found that heavier hydrocarbon components are more likely to stay in the adsorbents [17]. The adsorption of methane in activated carbon was highly sensitive to the pore size. When both micropores and mesopores were present, the maximum methane excess adsorption in the micropores (0.4 nm) was ten times (about 12.5 times) higher than that in the mesopores (9 nm) [18]. These molecular simulations refine the results of studies on the behavior of hydrocarbon components in porous media. However, the computation-intensive property limits the molecular scale simulations at a large scale (>100 nm). Simultaneously, the extensive application of CO 2 in unconventional oil/gas development brings new research topics. Zeng et al. [19] claimed that CO 2 did not form an apparent single adsorption layer in kerogen in organic nanopores. They pointed out that it was more reasonable to use the improved BET model for the adsorption amount. CO 2 has a robust competitive adsorption capacity for hydrocarbon components of shale oil. The selective adsorption capacity is comprehensively affected by reservoir temperature and pressure, kerogen type and maturity, water content, and pore surface wettability. Liu et al. [20] investigated the competitive adsorption of CO 2 -C 1 -C 4 mixtures in the double pore system with 1 nm and 3 nm of pore sizes (Figure 1(a)). The above results mainly focus on the competitive adsorption of CO 2 and lighter components in shale. Concerning the heavier components (n-alkane), simulation results indicated that CO 2 replaced the adsorbed hydrocarbons on the surface of calcite, and the amount of adsorbed CO 2 is controlled by the number of calcium potential on pore surfaces (Figure 1(b)) [21]. With the increase of CO 2 composition, CO 2 can mix with hydrocarbons in other 2 Geofluids pore areas only after the surface's adsorption layer is entirely saturated. Besides, temperature, pore size, CO 2 fraction, and length of n-alkanes are key factors in the selective adsorption of CO 2 and n-alkanes [21]. In general, the microscopic interaction mechanism between CO 2 and n-alkanes, such as octane and decane, has been rarely reported.

Phase
Behavior of Confined Fluids. Due to the significant portion of nanopores in such formations' total pores, the study of the influence of nanopores on the phase behavior of a single component and mixture is of great importance for improving oil and gas recovery in unconventional tight/shale reservoirs. As early as the 1960s, the effect of pore size on fluid properties has been known, and the relationship between phase behavior and pore size in porous media has been studied through physical experiments. Here, the results of phase behavior experiments over the past 60 years (Table 1) are summarized. It is found that the saturation pressure of pure components in nanopores is usually decreased significantly under a specific temperature. The critical temperature and pressure shifts occur with pore diameter decrease to the nanoscale. The hydrocarbon mixture phase sees a decrease in the bubble point pressure and upper dew point pressure and an increase in the lower dew point pressure. The shifts in vapor and liquid equilibrium (VLE) are also related to molecular chain length, wetting film, pore size, and microporosity. With the development of CO 2 -EOR and the increasing application of CO 2 storage in low pressure or depleted oil/gas field, more and more studies began to investigate capillary condensation of CO 2 -hydrocarbon mixtures confined in nanometer porous media.
Microfluidics visualization technology has advantages in speed, control, and sample size and can provide phase characterization analysis similar to existing batch methods. In contrast, nanofluids (<100 nm) provides unique insight into nanoscale phase transition phenomena. However, experiments often cost a lot to get inadequate data. Although pow-erful, imaging tools still do not understand the basic mechanisms by which fluids reside and flow in nanopores. Therefore, theoretical research has attracted more and more attention in recent years. Methods such as molecular dynamics (MD) simulation and density functional theory (DFT) have proved to be powerful tools for studying confined fluids' local properties. Grand canonical Monte Carlo simulation can be used to study the effects of confinement on adsorption, capillary condensation, and freezing/melting of fluids in nanopores [36]. Density functional theory (DFT) and mean-field approximation have been used to study the effects of nanoconfinement on fluids' thermodynamic properties. Based on the simulation results, a model was proposed to predict the adsorption of nonpolar and polar binary mixtures, and the relative contributions of fluid-wall and fluidfluid interactions were analyzed [37]. In applying nonlocal density functional theory (NLDFT), the pores of nitrogen and argon condense/evaporate in mesoporous molecular sieves (MMS) with cylindrical channels. The Lennard-Jones potential is used to consider the interaction potential between fluid and solid-fluid molecules. The prediction of adsorption/desorption isotherms by this model is in good agreement with the experimental data, but there is a significant deviation in pores less than 5 nm [38]. Using the Gibbs Ensemble Monte Carlo (GEMC) NVT instead of the traditional Grand Regular Monte Carlo Ensemble (GCMC) simulation to study the pure components (including methane, ethane, pentane, and n-decane) can avoid limitations to determine the phase transition location. In this way, a more accurate equilibrium pressure of liquid and gas can be obtained. GEMC-NPT-BPM (Bubble point Monte Carlo) was developed in conjunction with GEMC NVT simulations to obtain thermodynamic properties [39], including confined equilibrium pressure for the mixture. The simulation results show the shifts and closure of the confined fluid phase envelope relative to the bulk fluid. The critical temperature and pressure transfer to lower than the bulk value, the bubble   [20] and inorganic (b) [21] pores. 3 Geofluids point pressure decreases, and the dew point pressure increases. The adsorption of heavier components in methane/ethane and ethane/n-pentane mixtures increased in the vapor phase due to the nanopore walls' effect, and correspondingly, the adsorption of lighter components increased in the liquid phase. However, their applicability is limited to more complex problems, such as the spatial distribution of fluid mixtures in heterogeneous porous matrices due to their computational complexity.

Modified EOS and VLE Calculation.
It is impossible to study the distribution of fluid mixture in large-scale heterogeneous porous matrix by molecular-scale calculation. Therefore, researchers have focused on developing analytical (semiempirical) EOS models. An equation of state (EOS) in thermodynamic is an equation that provides the relations between temperature, pressure, volume, and some other variables. There are many different forms of cubic EOS in the literature. The van der Waals, back in 1873, introduced the first cubic EOS that was derived by the assumption of a finite volume occupied by the constituent molecules [40]. Based on van der Waals EOS, many other cubic EOS were derived [41][42][43][44][45], among which the Peng-Robinson (PR) EOS and the Soave-Redlich-Kwong (SRK) EOS are the most two widely used forms of EOS in the petroleum industry. However, although good prediction results can be obtained under specific temperature and pressure conditions, these cubic EOS are all in the assumption that the fluids are in bulk systems without boundary effects or restrictions. To better predict the hydrocarbon phase behavior in porous media, researchers suggest the conventional EOS should be modified to consider the wall forces' existence and the competition between the fluid-fluid and fluid-wall interactions of confined fluids.
In some of the earliest work, the researchers coupled capillary pressure effects to two-phase equilibria of hydrocarbon mixtures to meet unconventional reservoirs' nonnegligible capillary forces [46]. Tsallis [47,48] proposed to extend vdW EOS to predict critical temperature shifts with pore size qualitatively and quantitatively. Some subsequent studies apply vdW equations to the fluid calculation in slit-like pores based on perturbation theory [49]. In the calculation using the proposed new equation of state (EOS) incorporating capillary pressure, both bubble and dew point pressures are compared to experimental and molecular simulation results. Moreover, there is a maximum capillary pressure at each temperature beyond which equilibrium cannot be reached [50]. However, most of the predictions are not ideal when After the bulk liquid is completely evaporated, confined fluids then evaporate at a higher temperature or at a lower pressure than that in bulk space. 4 Geofluids approaching the critical point. Equation (1) describes twophase flash evaporation under capillary pressure.
Apart from merely adding capillary pressure, researchers have obtained the critical shifts of confined fluids by combining molecular simulation, DFT, and experimental data. One of the results of the investigation was the approximation described in Equation (2).
The effects of temperature on isothermal adsorption for Ar, N 2 , O 2 , C 2 H 4 , and CO 2 in mesoporous were investigated. Moreover, the quantitative relationship between pore size, molecular properties, and criticalities shifts was characterized [51,52]. The McM-41 sieves are set with the various pore. However, the study is based on classical thermodynamics and ignores the interaction between molecules and pore walls. Moreover, although the simulation results are consistent with the measurement results of capillary condensation, the influence of scale effect on the change process is not well reflected from the change of volume phase to the restricted critical property. Based on the generalized Van der Waals equation of state theory for confined Lennard-Jones fluids in nanopores, a new variation estimation formula for critical attributes is proposed, as shown in Equation (3) [53]. In this proposed correlation, the adsorption and desorption in nanopores are ignored.
A configurational-bias grand-canonical transitionmatrix Monte Carlo simulation was carried out for the critical temperature in smaller nanopores (0.5-5 nm) [54]. This simulation result is summarized as the general correlation between ΔT c and d p /σ LJ , as shown in Equation (4). This correlation is widely used in EOS models with shifted critical properties in nanoscale porous media [55][56][57][58][59]. However, lower critical pressures were observed in the following molecular simulation of Lennard-Jones fluids in carbon nanopores with adsorption properties [60,61].
ΔT c = 0:6, when d p σ LJ < 1:5, Singh's results were subsequently used to investigate the influence of molecular weight on the critical properties of confined single-component hydrocarbons [56] and the specific changes in phase behavior and critical properties of different hydrocarbon mixtures in the nanospaces [62,63]. Later, Islam et al. [64] compared the difference in confined phase behavior description within Peng-Robinson EOS and the vdW based on a new proposed correlation. The prediction results of this new relationship are in good agreement with experiments and molecular simulation in the whole range of pore sizes from the scale of single molecule (σ LJ /r p ≤ 0:5) to the bulk spaces. The critical terms obtained are Furthermore, Travalloni et al. [65] modified PR-EOS (PR-C EOS) by adding terms related to pore size into the equation to influence molecule-wall interactions. Pore size and shape were taken into account in the phase equilibrium calculation. This method can consider the influence of molecular-solid wall interaction energy in different pore sizes and matrix types (heterogeneous adsorbent) when predicting the phase behavior in bulk and confined space.
x i x j a p,ij À Á , In the following research, the above method has got more application and promotion. For example, Ibrahim [66] used PRC-EOS to study the phase behavior of fluids in pores of different structures with pore sizes ranging from 2 nm to 5 nm. The condensation phenomena in different pore sizes are verified, and the phase behavior in porous media is predicted. Luo et al. [67,68] studied the mixtures' phase equilibrium after injecting methane into the Anadarko Basin shale oil. In the physical model applied in this study, the pore size distribution is discretized into specific pore sizes, which represent a multiscale system of typical pore size distribution. The simulation results show that the nanopores can reduce the fluid pressure below the bubble point, thus making the mixture subcritical. The deviation of bubble point pressure is also related to the amount of injected gas apart from pore size. The phase behavior shift in nanopores tends to contain more swelled fluid than that in the bulk state.

Interfacial Tension and Miscibility.
Recent years have seen an increase in the application of CO 2 injection to enhance unconventional oil/gas recovery worldwide. This trend is due to the unique properties of CO 2 , including higher solubility in crude oil, lower minimum miscible pressure, larger diffusion coefficient in nanoporous media, and higher Langmuir adsorption in shale organic matrix.
Interfacial tension (IFT) significantly affects the flow trajectory, gas-liquid equilibrium, and saturation distribution of fluids in porous media due to the cumulative contribution of Jammin effects in pores and throats to flow resistance. Also, since CO 2 -brine IFT is less than that of hydrocarbon-brine, the breakthrough pressure of CO 2 may be much less than that of in situ reservoir gas [69]. Therefore, to ensure seal integrity, the CO 2 injection and storage pressure must be lower than the initial reservoir gas pressure. Moreover, the CO 2 injection pressure threshold must be predicted based on the IFT of the fluid in the specific reservoir. In other words, the IFT behavior of gas-liquid systems in unconventional reservoirs significantly affects the transport characteristics of CO 2 and the efficiency of capillary sealing. Therefore, achieving optimal CO 2 flooding and sequestration requires an accurate assessment of the IFT in the system.
Numerous experimental and simulation studies have been conducted to quantify the IFT of various gas-water systems. The most widely used IFT measurement technique is Axisymmetric Drop Shape Analysis (ADSA) method. The drops' geometric shape can be measured by the ADSA method, and then IFT can be calculated by the young-Laplace equation [70]. At present, extensive experimental measurements of IFT in gas-water systems have been made over a wide range of temperatures and pressures [70]. Chao et al. [71] explored the effective pore throat size from the perspective of the nonnegligible value of gas-water capillary force (Figure 2(a)). However, most studies have not considered the effect of nonhydrocarbon gas pollutants on system IFT, especially under high temperature/pressure conditions in reservoirs. Besides, pure water without salt is contained in most systems [72]. It has been recognized that the salinity of water does significantly affect IFT. One explanation holds that the adsorption of cations on the interface leads to the accumulation of cations in the aqueous phase. Hence, the effect of salinity on IFT is due to the interface structure changes. Another explanation is that IFT increases as the density of the aqueous increases due to the dissolved salt ions [73,74]. Many studies listed the IFT of CO 2 -water and CO 2brine systems, and the measured temperature and pressure range were as wide as~290-470 K and~2-100 MPa, respectively. Factor analysis also involved the effects of salt concentration (0.085-2.75 mol/kg) and types (NaCl and CaCl 2 ) [75][76][77][78].
Besides, assessing the extent of structure and residual capture also requires a better understanding of the interfacial properties of CO 2 with in situ fluid and mineral systems. Although there is much uncertainty in the data, the CO 2 critical points measured at high pressures are very similar. Simultaneously, several similar conclusions have been found in many experimental studies: (1) in the carbon dioxide subcritical zone, IFT decreases sharply with pressure. In contrast, in the supercritical zone, IFT reaches a platform. (2) IFT usually decreases with the increase of temperature, but it reacts to more temperature dependence when CO 2 is near the critical point [79]. (3) IFT increases linearly with salt concentration [74]. Parachor Model and Scaling Law have been more widely used in IFT prediction due to their simplicity, but they cannot perform well for hydrogen-water systems. Other IFT prediction models have been reported in some studies, among which the Firoozabadi and Ramey model [80] can be used to predict IFT of oil/gas and water mixtures. Based on this, some scholars proposed a new IFT correlation by adding parameters such as compound molar mass for correction [81,82], but their predictive ability is still limited. In the correlation of Bachu and Bennion [73], IFT is a function of salinity. This model can predict the IFT of the system based on the solubility of CO 2 in brine. However, this model is not applicable to predict IFT in the reservoir with a high temperature (>398 K) and pressure (>27 MPa). Besides, this kind of correlation can only consider the difference of temperature, pressure, and density of pure components, but it is challenging to analyze the influence of mutual solubility [78,83].
Classical molecular dynamics (MD) has been used to predict the interfacial tension between CO 2 and water (brine) [84], as well as the CO 2 -water-mineral surface contact angle [85]. The latter has been proved to be a function of temperature, pressure, and salinity. The IFT prediction accuracy depends only on whether the selected force field can reproduce the interfacial energy. However, MD simulations are limited to nanoscale studies when determining contact angles, which are smaller than most macroscopic measurements of droplet size (millimeter level). The influence of the contact line's curvature on the contact angle can be ignored in the macroscopic size of the droplet. However, at the nanoscale, the liquid and solid surface forces will change the droplet's shape, resulting in the influence of the contact line's curvature that cannot be ignored [85,86]. The influence of contact lines is usually described by introducing contact line tension [87]. Other statistical thermodynamic methods, such as linear-gradient theory, perturbation theory, density gradient theory (DGT), integral and density 6 Geofluids functional theory (DFT), are also used to predict IFT. As a representative of the silicate strata' composition, quartz and mica have been used as the mineral base for contact angle measurement in many experiments. Some studies have also reported gas-liquid contact angles on the calcite surface (carbonate formation) [85,[88][89][90]. Since calcite is a suitable material to adsorb organic matters even in dilute solutions and organic materials and adsorbed ions can significantly change the wettability of mineral surfaces, the contact angle can be observed from the initial strong water wet (0°< θ < 75°), scattering to between 2°and 60° [91]. Atomic force microscopy (AFM) data and simulations show that the adsorbed ions behave as ordered water layers on the calcite surface rather than directly binding to mineral atoms [92][93][94]. Also, calcite's surface properties depend mainly on the history of its exposure-CO 2 increases the acidity of water droplets, which causes calcite to dissolve, changing the local surface morphology [95].
In an oil-bearing reservoir, the injected CO 2 has two-way mass transfer with the in situ oil. CO 2 is dissolved in the oil, and hydrocarbon components in the in situ oil are extracted, and then the new mixtures are formed and further transported. As the composition of the remaining oil changes, so does the IFT of the gas-liquid system. Therefore, the initial fluid composition is considered one of the essential factors affecting IFT [96][97][98]. Also, there is a controversial view that the initial gas (CO 2 ) fraction will also have a nonnegligible influence on IFT. Earlier studies only confirmed that CO 2 fraction would affect the speed of reaching gas-liquid equilibrium [96], and later studies found that IFT also increased slightly when the initial gas-oil ratio (GOR) increased [98]. In general, IFT is positively correlated with temperature,  7 Geofluids pressure, initial fluid composition, and feed gas to liquid ratio (FGLR). Besides, under specific temperature and pressure conditions, the system IFT after CO 2 injection can reach zero. That is, the system is entirely miscible [99]. Simultaneously, when the test pressure is higher than the mixed system's saturation pressure, zero IFT can also be observed [100]. How to distinguish clearly between miscibility and saturation needs further study.
In recent years, fluid characteristics confined in micropores and even nanopores have attracted more and more attention in practical applications. Current methods for IFT calculation and assessment of influencing factors include improved EOS equations and molecular simulations [101][102][103][104][105]. The general conclusion is that a decrease in pore size leads to a reduction in IFT (Figure 2(a)) [106]. The minimum miscible pressure (MMP) of CO 2 and in situ oil is an inevitable problem in the study of CO 2 -EOR. As described above, apart from pore sizes, IFT is also affected by reservoir temperature, hydrocarbon composition, and initial gas-oil ratio. Therefore, these factors are also important factors affecting MMP [107]. The commonly used MMP calculation methods, including the method of characteristics (MOC), multiple mixed cells (MMC), and slim tube simulation method. However, the system MMPs in nanopores need to be calculated by combining with the modified EOS or the Perturbed-Chain Statistical Associating Fluid Theory (PC-SAFT) EOS and the Vanishing Interfacial tension (VIT) method [108]. Some calculations have found that changes in the critical properties lead to a decrease in MMP, while capillary pressure has no effect on MMP [104]. This phenomenon is mainly due to two pseudocomponents with the same critical point in the mixed system used for calculation. The IFT reaches zero after the critical point is exceeded. It has also been found that in the immiscible region (where the pressure is lower than MMP), high capillary pressure can improve the recovery efficiency of heavier components and can change MMP. However, MMP changes inconsistently with increasing capillary pressure and may decrease, increase, or remain unchanged for different mixtures. When both capillary force and critical shifts are considered, the MMP of the mixture is reduced by nanopores, but when the pore diameter is greater than 10 nm, the change of MMP can be ignored [109,110]. In immiscible displacement, the capillary pressure effect is useful for enhancing oil recovery. However, for the miscible or near-miscible situations, the influence of capillary pressure on fluid dynamics will be compromised [111].

Micropore Structure Changes with CO 2 Injection
3.1. Mineral Dissolution and Deposition. Accurate prediction of petrophysical properties, the integrity of CO 2 storage seals, and flow characteristics require a full understanding of the reaction transport process [112,113]. The reactions in this process occur between fluid-fluid (homogeneous) and fluidsolid (heterogeneous). Specifically, it can be summarized by the following two types of reactions.

C aq ð Þ + M s ð Þ = P aq ð Þ: ð7Þ
Dissolved CO 2 acidifies water through the following reaction: The dissolution of CO 2 decreases the solution's pH and accelerates the dissolution of minerals in the host rocks. Further water-mineral reactions provide divalent cations in solution. For example, calcium-bearing plagioclase (such as anorthite) dissolution can provide Ca 2+ . Moreover, olivine dissolution can provide Fe 2+ and Mg 2+ .
Mineral trapping is considered the most permanent form of geologic carbon storage [114]. Mineral carbonation reactions are thermodynamically favored but proceed slowly [115]. Most minerals in sandstone reservoirs are not reactive for carbonation reactions, so it may take hundreds to thousands of years for CO 2 mineral trapping to happen [116]. Formations containing richer Ca, Mg, and Fe can potentially result in mineral trapping over much shorter time scales.
In the case of an evolving porous medium affected by mineral dissolution-precipitation reactions, it is evident that both porosity and connectivity can be affected. The dissolution of a solid phase is directly linked to the increase of the overall porosity. However, dissolution does not necessarily increase effective porosity to the same degree. The effect can be substantial if dissolution widens and opens up pore throats or opens up new diffusion pathways [117,118]. It can be more limited if dissolution is more equally distributed [119]. Similarly, mineral precipitation might have a limited effect on connectivity if minerals form on existing surfaces without clogging pore throats.
On the other hand, precipitation may lead to the disconnection of existing solute migration pathways [119], with a pronounced impact on flow and transport parameters [120][121][122]. These theoretical considerations imply the dissolution, and precipitation of minerals can lead to very different impacts on effective porosity and connectivity with the same total porosity change. The porous media's total porosity includes transport pores φ tra , dead-end pores, and isolated 8 Geofluids pores. Apart from the isolated pores where the fluids cannot directly contact injected gas, mass transfer between CO 2 and in situ fluids occurs in the open pores φ open (including transport pores and dead-end pores). The dissolution/precipitation caused by chemical reactions may penetrate or increase some of the isolated pores and change the critical radius r cr and tortuosity τ that affecting bubbles to pass through. Therefore, the total porosity, effective porosity, and porethroat connectivity have a combined effect on tortuosity and permeability, affecting flow and migration ( Figure 3) [123,124]. After the acid fluid comes into contact with the mineral, the dissolution occurs in the undersaturated state, increasing fracture aperture. Mineral precipitation may occur when the concentration of necessary hydrates reaches a certain threshold after mineral dissolution [125][126][127]. For example, gypsum may precipitate after calcite is dissolved [128,129], and calcite may precipitate after dolomite is dissolved. In these cases, fracture permeability change direction and extent depend on the interplay between dissolution and precipitation.
In the early stage, some short-term reaction experiments related to CO 2 injection were proposed. Rock samples with different permeabilities were prepared by hot pressing of calcite powder and quartz powder, and then CO 2 was injected into the samples. Based on micro-CT imaging, pore microstructure changes during limestone dissolution can be observed, and the power-law relationship between permeability and porosity can be proposed [130,131]. However, in the macroscopic reservoir, the solute is not evenly distributed in the heterogeneous porous medium. This feature may lead to a deviation in the estimation of the reaction rate [132]. By comparing the Poiseuille flow model with a well-mixed reactor model, Li et al. [133] studied the rate of mineral reactions in a single pore and fracture and discussed the physical model scale's impact. In short-term experiments on Tensleep sandstone and Berea sandstone (mainly dolomite and anhydrite cement), core samples were washed by CO 2 for several days after saturated salt solution. It was found that the precipitation of minerals led to a decrease in permeability despite the presence of mineral dissolution [134,135]. Another experiment on sandstone samples cemented with calcite and dolomite showed that the dissolution of carbonate minerals led to an increase in permeability [136]. This increase is due to enhanced pore modification and is related to the dissolution of carbonate cement bonds. In the long run, minerals in the formation that react with CO 2 in brine include calcite, siderite, dolomite, quartz, barite, muscovite, feldspar, and clay minerals [137,138]. Chemical reactions will dissolve minerals, weaken cementation, weaken particle time bonding, and alter flow characteristics, thereby affecting CO 2 displacement and storage operations' efficiency and safety.
Given the complexity and heterogeneity of pore structures, extensive chemical disturbances, and the coupling between these different processes, it is difficult to obtain quantitative assessments of dissolution patterns and rates and feedback on porosity and permeability based solely on theoretical considerations. Recent advancements in imaging techniques, such as X-ray microtomography [139] and nuclear magnetic resonance (NMR) [140], allow visualization of pore structures at submicron resolutions, providing a window into the complex properties of porous media. Simultaneously, various methods ranging from traditional computational fluid dynamics techniques to smooth particle hydrodynamics and lattice Boltzmann methods have been proposed to simulate the flow of porous media [141][142][143][144]. With improved imaging technology combined with the development of pore-scale model formulations [123,145,146] and substantial advances in computational power, the exploration of processes on the pore scale has become possible from both an experimental and modeling perspective [147][148][149][150].
The pore network model comprises interconnected throats and pores, reflecting the complex structure of porous media. Many scholars use the pore network model to characterize the physical and mineral properties of porous media. Li et al. [151] and Raoof et al. [152] studied the chemical reaction rate in porous media by using the pore network model and established the relationship between the Peclet number and the adhesion/separation coefficient of adsorbents. The pore network model has also been used to study chemical reactions on porosity, permeability, and the correlation between them in the homogeneous system [119]. Tansey et al. [153] proposed a mass transfer coefficient correlation and pore-merging criterion by simulating carbonate rocks' dissolution. The pore-scale model and the direct simulation based on finite-difference were used to solve the Stokes equation and convection-diffusion equation. However, the comparison of the results showed that the pore network model was better adapted to the case of larger velocity and reaction rate under the same situation [154]. Besides, the pore network model also has some limitations. For example, it is difficult or sometimes impossible to extract representative pore network models for heterogeneous porous media [155].
A Lattice Boltzmann method (LBM) was developed to simulate the porosity scale dissolution and precipitation [156]. The method can simulate the multicomponent reaction transport process under different temperatures and pressures. Also, LBM can be developed to simulate reactive multiphase flows in arbitrary geometric pores. The temporal and spatial distribution of each constituent phase and the reactants' evolution during the dissolution/precipitation process are reflected [157]. As mentioned earlier, the nonuniform distribution of reactants in the heterogeneous pore structure will result in differences in the local reaction rate.
On the one hand, the researchers established a pore-scale numerical model based on LBM. On the other hand, a continuous scale numerical model directly simulated is also established. By comparing the results of the two simulations, it was found that the inhomogeneity of the flow field would reduce the reactivity of the system. The more substantial the pore structure's heterogeneity was, the more pronounced the reduction was, leading to the difference in the velocity of the reaction front between different models [112,158]. Combining the LBM method with the flow-solid interface tracking method can predict the positive and negative effects of mineral dissolution/precipitation on porous media [159] and the relationship between permeability and porosity [160]. The numerical results' complexity is reflected in the fact that the correlation between pore and permeability is highly dependent on Peclet and Damkohler systems and the porous medium structure. Extending the LBM method, the effects of decomposition, dissolution, and precipitation in fluid-mineral systems on multiphase transport can be discussed in depth [161].

Asphaltene Precipitation.
The precipitation and deposition of long-chain alkanes (mainly asphaltenes) in conventional reservoirs' pore space have been studied extensively. Researchers define phase precipitation as a solid phase transformed by a liquid phase, whereas deposition refers to a precipitate (solid phase) attachment to a rock surface (Figure 4) [162]. The current asphaltene precipitation thermodynamic models assume including colloid-based methods and solubility-based methods. The model based on the colloidal approach assumes that asphaltenes are suspended solid particles, which are colloidal in the system. Physical and chemical reactions disrupt the thermodynamic equilibrium, which may cause the outer layer rupture and asphaltene flocculation [163]. The method assumes that the asphaltene components based on solubility can be decomposed into the components and settling components, the precipitation components can dissolve in the oil and form a solution, and precipitation occurred in solid-liquid equilibrium (SLE) or liquid-liquid equilibrium (LLE) state, can through the SLE solubility model and another physical model to forecast the asphaltene flocculation [164]. There are three types of asphaltene precipitation theoretical studies: using the Support Vector Machine (SVM) model to study the development of asphaltene precipitation [165]; applying the Computational Fluid Dynamics Method to investigate the molecular deposition of asphaltic materials [166,167]; and using molecular simulations to understand asphaltene deposition behavior by studying interactions between different components of molecules [168].
In the process of gas injection for oil recovery, the injected gases change the inherent dispersion system of col-loids and asphaltenes in crude oil, leading to instability of asphaltenes, which in turn leads to viscous solid deposition [169]. On the one hand, asphaltenes are adsorbed on reservoir rocks, blocking pore throats or changing formation wettability, reducing hydrocarbon fluids' mobility [170]. On the other hand, asphaltene deposition can cause severe formation damage to reservoirs, especially near production wells with the most remarkable pressure drop. Asphaltenes can also flocculate on rock surfaces. Flocs rupture during dissociation and plug the pore openings of the reservoir [162]. Different injection gases have different effects on asphaltene precipitation. The injection of CH 4 , CO 2 , and N 2 will lead to asphaltene precipitation, among which N 2 has the most significant influence on asphaltene precipitation [171,172], followed by CH 4 [173].
According to the currently known, there are three main reasons for the asphaltenes caused by CO 2 injection. The first one is the changes in temperature and pressure [174,175]. Studies have confirmed that the temperature drop caused by CO 2 injection in the near-well zone would result in asphaltene precipitation [174]. The size of asphaltene aggregates in CO 2 is negatively correlated with temperature, and the nonmonotonic effect of pressure on aggregates can also be observed. Asphaltene deposition reaches its peak value at the oil bubble point [176]. Also, resins or other polymers' presence limits the tendency of asphaltenes to accumulate in CO 2 [177][178][179]. When the fluid pressure exceeds MMP, asphaltenes begin to precipitation. When the system pressure is below the MMP, the pore blockage damage is much less than when the pressure is higher than MMP. This is because the resins that stabilize asphaltenes are more unstable when pressure is higher than MMP [180][181][182][183]. The second reason is CO 2 concentration [184]. Some other studies have shown that the flocculation of asphalts in single-phase mixtures of petroleum and carbon dioxide has nothing to do with pressure but only with carbon dioxide concentration. They also found that asphaltene flocculation increased linearly with the increase of CO 2 concentration [185]. Compared with heptane and toluene, asphaltene stable aggregates in supercritical carbon dioxide have a longer lifetime [186]. Wang and Ferguson [187] used the force field to establish a mesoscale method for studying asphaltene aggregation. A flexible structure is used to deal with carbon dioxide. Visual molecular dynamics (VMD) was used to prepare visualizations [188]. Mehana et al. studied the system's physical conditions and composition on asphaltene aggregation dynamics [189]. The third reason is separation and selective extraction by CO 2 [190]. When the lighter components are extracted and carried away by CO 2 , the remaining hydrocarbon mixtures' molecular weight gradually increases, increasing the risk of asphaltene precipitation and formation damage. What is more, asphaltene precipitation is closely related to the composition of rocks and minerals. Calcite content is positively correlated with asphaltene deposition [175], while iron content is negatively correlated with asphaltene deposition [191]. Most asphaltene crude oils have increased wettability on mica surface [192]. Due to the higher wettability of reservoir crude oil, the content of bound water in pores decreases (from 26.5% to 10.7%), leading to increased water production and decreased final oil recovery [193]. The presence of water will lead to a reduction of asphaltene precipitation [194]. This reduction is due to the buffering properties of the formation brine.
At present, there are little researches on how asphaltenes in crude oil behave in micropores and nanopores during CO 2 injection. Mohammad et al. [195] simulated asphaltene deposition and studied the optimal CO 2 injection during saline-CO 2 cycle displacement to improve the recovery of low permeability reservoirs by CO 2 flooding. Shen and Sheng [196] studied asphaltene deposition produced by cyclic CO 2 injection into the Eagle Ford Shale. However, this study was conducted at room temperature (~25°C) and 0.3 MPa pressure, and no mechanism of asphaltene precipitation, deposition, and plugging in nanoporous has been reported. Fakher and Imqam [197] found that because unconventional reservoirs' pore size is close to the size of asphaltene clusters, the asphaltene is difficult to extrude from pores. A large amount of asphaltene accumulates in the nanopore and blocks the rock matrix severely. Sun et al. [198] studied the competitive adsorption of asphaltenes by carbon dioxide and nitrogen in nanopores to study the feasibility of using asphaltenes as a means of carbon capture in nanopores during carbon dioxide injection.

Stress Field Redistribution.
Although CO 2 has long been injected into depleting reservoirs to enhance oil and gas recovery, there is a relatively little field experience in conducting geological CO 2 storage in oil layers than the mature and commercially viable technology of pumping CO 2 underground. A significant difference between CO 2 -EOR and CCS is the cost of CO 2 as a displacing agent. EOR is intended to make a profit by reducing the injection of CO 2 and reuse it after being produced as a dissolved gas. However, the goal of CCS is to maximize the storage of CO 2 and permanently trap it in place. In the past few decades, the scientific literature has discussed issues related to the sequestration of CO 2 underground from the perspectives of hydrodynamics, geochemistry, and geomechanics [199][200][201][202][203][204].
Pore pressure buildup induced by CO 2 injection depends on formation capacity, transport properties, and aquifer connectivity [205,206]. Formation transport and geomechanical properties, including porosity, permeability, and pore compressibility, determine CO 2 plume movement and local pore pressure changes. These formation properties vary considerably with rock types and spatial heterogeneity, significantly impacting CO 2 trapping and migration. Pressure changes induced by CO 2 injection operations alter stress at the geological carbon storage (GCS) site. For example, changes in geomechanical conditions may result in mechanical failure of the reservoir or reservoir expansion. Stress arching may increase the vertical stress over the reservoir and lead to normal fault activation. Intricate stress transfer patterns between reservoir and caprock, associated with the lithologies' poroelastic properties and the reservoir stress path resulting from the injection, could result in tensile or shear failures and enhanced fluid flow. These reservoir mechanics changes caused by stress field redistribution may also provide pathways for CO 2 leakage or trigger induced seismicity. Maintaining CO 2 injection for several years will lead to a large pressurization area within the reservoir [207]. Even if the CO 2 diffusion radius is only a few kilometers, the pressure perturbation cone's radius with overpressure higher than 0.1 MPa can exceed several hundred kilometers [208]. Because of the similarities between CO 2 injection and other subsurface activities involving fluid injection, the seismic risk associated with carbon storage operations is real, and risk assessment is critical to informing any decision-making from site characterization to the determination of operational parameters.
Some researches suggested that the high pore pressure could cause hydraulic fractures in the shale caprock [209,210]. The fractures or cracks generate the structural environment (including squeeze, stretch or slip), stress difference, and the amount and orientation of brittle ruptures [211]. These stress characteristics describe the leak-off of oil and gas through expanding fractures developed in overpressure reservoirs. In this case, the fractures can reseal when the formation returns to its normal state of pressure [210]. To prevent the formation of these cracks or fissures, the researchers propose limiting the injection pressure below the minimum stress (the minimum pressure that causes tensile fractures).
However, some studies [211] suggest that tensile cracking can occur only under relatively low-stress differential conditions. In their view, the stress caused by reservoir expansion in the caprock is not the cause of the fault generation, despite the possibility of shear failure at the reservoir-cap interface and hydraulic fractures extending from the reservoir to the caprock above. At the same time, they emphasize that reactivation of existing faults is a greater risk, or even one of the significant risks, of triggering cap leakage ( Figure 5) [212]. Changes in the in situ stress conditions associated with  11 Geofluids subsurface activities can cause suitably oriented faults or fractures to slip because of shear. Fractures that are oriented to maximize the possibility of slip are referred to as "critically oriented" fractures.
Similarly, fractures that would require only a minor increase in fluid pressure to slip are referred to as "critically stressed" fractures. Higher pressure cutoffs that may induce fault stability issues in critically stressed faults, such as 1 MPa, can extend tens of kilometers [208,213]. Furthermore, once large-scale geologic carbon storage projects are fully deployed, the superposition of overpressure from different injection wells in the same sedimentary basin is expectable, leading to an even larger pressurized region [214]. Thus, even though injection wells will most likely be placed far away from identified faults to minimize the risk of fault reactivation, overpressure will eventually reach faults in the far field, affecting their stability [215]. For any activities involving fluid injection, evaluating critically stressed fractures is fundamental for understanding the placement and orientation of injecting wells, determining the main operational parameters such as the injection rate and maximum allowable injection pressure allowed, or deploying a riskbased monitoring strategy.
On the other hand, the variations in pore structures due to reaction have a significant impact not only on petrophysical properties but also on the porous media's mechanical properties. In reactive transport, porosity changes due to reaction. This results in significant variations in the porous media's mechanical properties [216].
Stress and strain changes depend on the number of reactants and the amount of dissolution and precipitation produced in the pore structure. Specifically, due to the lack of active minerals, only minor chemical-mechanical changes are expected in most quartz-sandstone and unconsolidated sandstone when exposed to CO 2 . In terms of geomechanical property types, sandstone with "CO 2 -weak" intergranular cement is the most easily changed rock. Even a small amount of reaction can significantly impact the geomechanical properties that depend on cementation. For some types of rock, including carbonate-cemented sandstones (e.g., Entrada sandstone and Castlegate sandstone) and clay-cemented sandstones (Chlorite-cemented Tuscaloosa sandstone and Mt. Simon sandstone) [217], dissolution and degradation of the pressure consolidation will result in (1) decreases of cohesive strength, fracture toughness, and yield stress locus size; (2) increases of compliance and creep; (3) changes in postpeak behavior; and (4) changes in frictional behavior [217][218][219][220][221]. When granular-supported carbonate rocks (e.g., chalk) are exposed to acidified (dissolved CO 2 ) brines, significant stresses occur [222]. Pores in carbonate rocks will expand significantly, but the rocks' mechanical properties will not change much until a large number of rock minerals are dissolved [223]. In carbonate-rich shale, soluble minerals are distributed in sheets between soluble bedrock and cement sandstone [224]. However, due to the difficulties in sample processing and pore structure protection, core-scale CO 2 and shale cap reaction experiments are limited [225].
Chemical-stress interactions in the reservoir can lead to wellbore stability issues [113] and CO 2 storage safety risks.
First, chemical reactions that dissolve minerals can cause rock damage, increasing the risk of wellbore collapse. Simultaneously, the reaction of CO 2 with minerals reduces the mechanical strength of the formation containing brine, which may lead to CO 2 leakage [226]. Therefore, it is crucial to understand the influence of reaction flow on mechanical rock properties in CCUS. Madland et al. [227] studied the mechanical stability of chalk after saturated by CO 2 gas and carbonic acid water. The results show that the influence of CO 2 gas on the strength of chalk is small, while that of carbonated water is great. Farquhar et al. [228] tested the mineral and porosity changes of sandstone and siltstone after reacting with low-salinity water and supercritical carbon dioxide (SC-CO 2 ). The test results showed that the calcite content decreased from 17 vol % to 15 vol % after the reaction, and the porosity increased by 1.1 vol %. CO 2 -water-rock interaction also causes cap deformation, leading to significant stress and stress-induced permeability changes [229,230]. Few studies have focused on the effect of the reaction on porous media's mechanical properties, and the design of CO 2 injection schemes for unconventional reservoirs often ignores the strain and stress caused by chemistry.

Challenges and Suggestions
Unconventional reservoirs are characterized by organic and inorganic matters, diverse mineral compositions, nonignorable capillary forces, and phase state changes under micronanopores' confinement [231]. The surface wettability and adsorption characteristics of different minerals present great complexity. Therefore, incorporating the processes that occur inside mineral nanopores and recognize the flow characteristics of CO 2 in the matrix and natural-hydraulic fracture system is extremely difficult.
Due to the dense matrix rock, injected CO 2 first passes quickly through the fracture. Then, affected by the pressure gradient, CO 2 seeps into the matrix and replaces the in situ oil in it. As more crude oil enters the fracture and the bulk CO 2 , the injected CO 2 permeates into the matrix. After that, the oil in the bulk CO 2 is driven out due to the expansion and viscosity decrease. Finally, the dominating driving force changes to concentration gradient diffusion from matrix to fracture with the continuing decline pressure gradient. In the current studies, the nanopore confinement affects the molecular density distribution and phase behavior, which results in the difference of gas-liquid equilibrium, interfacial tension, capillary force, and miscibility from that in conventional millimeter porous media [232]. Simultaneously, mineral dissolution and precipitation, asphaltene precipitation, and stress field changes caused by chemical/physical reactions between fluids and rocks may change the geometric characteristics and connectivity of pores, affecting the apparent permeability of fluids.
On the other hand, after CO 2 is injected into the pores of shale containing water or oil, the rock surface's wettability will significantly change, further affecting the microscopic migration mechanism of fluid components in the reservoir. The film formed by CO 2 adsorption can transform the local hydrophilic surface in kerogen pores into a superhydrophobic 12 Geofluids surface on the surface of organic matter, reducing the flow resistance of formation water [233]. On the surface of inorganic matter, CO 2 can peel off the nonpolar components of in situ oil, leaving the polar components, and enhance the hydrophilicity by selective extraction [234]. In the formation of water, CO 2 , and pore surface, CO 2 can make mica and quartz surfaces with strong water wettability behave as neutral wetting. This phenomenon is significantly affected by mineral concentration [235]. Although it is not apparent, the boundary slip exists in the oil phase displaced by CO 2 in the inorganic pores. The flow of formation water will change from nonslip boundary to slip boundary on the kerogen pore surface after forming an absorption layer of CO 2 [236]. The change of wettability leads to more complicated diffusion and migration rules of multiphase fluid confined to the nanopore. Fluid migration in the subsurface can induce changes in phase state, mechanical stresses, and the fluid and natural geological material chemistry. In many situations, changes in all processes give a coupled response, in that one process feeds back to another. When trying to understand CO 2 -oil flow through nanoscale porous media, it is essential to identify the relative importance of the processes occurring and the degree of interactions. Modeling of such highly nonlinear coupled flow should be involved, but current and predicted computational ability could not simulate all the known and physically described processes operating efficiently [237][238][239]. Although CO 2 injection has been widely used in conventional reservoirs to enhance oil and gas recovery, the interaction of CO 2 with fluids and pore surface in unconventional formations will influence phase behavior and multiphase flow, thus affecting CO 2 capture and reservoir stress redistribution. However, the current research results cannot reasonably explain and describe these changes. Most largescale simulations of CO 2 storage intrinsically neglect the complex and vital interaction between CO 2 and multicomponent mobile liquid hydrocarbons.

Conclusions
This review summarizes some fluid-fluid and fluid-solid systems' physical and chemical reactions with CO 2 injection into unconventional reservoirs. Also, the resultant change of pore structure and connectivity is discussed in this paper. Regarding the limitation recently, the challenges and some suggestions of CCUS for the future are proposed. Based on this review, the following conclusions are drawn below.
(1) It is a practical experimental method to study the phase behavior through microfluidic technology. Moreover, the visualization method of nanofluids can observe some unique phenomena of nanoscale phase transition. Theoretical approaches like MD and DFT can make up for high-cost defects and are limited to high temperature and pressure conditions in the microscopic phase experiment (2) The modified EOS can be used to deal with VLE calculation, including nanopore confinement effect, but further investigation is needed to predict the phase behavior for long-chain hydrocarbon mixture within a wide temperature and pressure range (3) Molecular simulation technology has been widely used to study the adsorption/desorption of CO 2 in micro-nanoscale pores. In recent years, we see the crescent studies of microscopic interaction between CO 2 and heavier oil components (4) Salinity and temperature affect interface properties of CO 2 -brine-mineral system. The specific phase behavior will change the IFT and alter the MMP in oilbearing nanopores (5) In addition to total porosity, mineral dissolution and precipitation of minerals may have a very different impact on effective porosity and connectivity.

Geofluids
Improved imaging techniques, combined with numerical methods, provide a solution for long-term pore geometry changes and permeability feedback (6) CO 2 extraction may increase the risk of asphaltene deposition, which may block pore throats or alter formation wettability. However, more researches are needed for the performance of asphaltenes in microns and nanopores after CO 2 injection (7) The mechanical properties of rocks are affected by CO 2 pressurization and the chemical reaction. The resulting stress changes can cause rock damage or pose a microseismic risk to the formation, opening microfractures, restarting faults, or creating artificial fractures (8) The study of CCUS operation and safety assessment in unconventional reservoirs requires the coupling of phase behavior, fluid and geological change, and stress field redistribution Nomenclature a p : van der Waals energy parameter modified by confinement b p : van der Waals volume parameter modified by confinement d p : Pore diameter F p : The fraction of the fluid molecules in the confined space f : Fugacity g s : Granular size N av : Avogadro's number N c : Index denotes the number of fluid components of the mixture P: Pressure P b : Bubble point pressure P cap : Capillary pressure P d : Dew point pressure r p : Pore radius T: Absolute temperature v: Molar volume of the fluid x: Liquid molar fraction y: Vapor molar fraction.

Data Availability
All the data is presented in the manuscript.

Conflicts of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.