1. Introduction
Anthropogenic carbon dioxide (CO
2) emissions, which account for over 75% of global greenhouse gases, are the primary driver of accelerating climate change
[1]. The Intergovernmental Panel on Climate Change warns that, without urgent action to reduce emissions—specifically a 43% reduction from 2019 levels by 2030—global temperatures could exceed 1.5 °C as early as 2030
[2]. Geological CO
2 storage is a critical component of climate mitigation strategies
[3],
[4], with the capacity to permanently sequester billions of tons of CO
2 and facilitate deep decarbonization. Sandstone formations are frequently targeted for CO
2 storage due to their widespread availability, structural stability, high porosity, and ability to securely trap CO
2. However, during CO
2 injection, the formation of acidic CO
2–saturated brine inevitably triggers reactions with the minerals in the sandstone
[5]. Robust findings from experiments and simulations by Cui et al.
[6], Luhmann et al.
[7], and Ma et al.
[8] demonstrate that these geochemical reactions can have a significant impact on key performance metrics, including the CO
2 injectivity, storage capacity, and long-term safety of the storage site.
Extensive research has been carried out on reactive flow during CO
2 injection into sandstone, with a principal focus on the effect of reactive flow on absolute permeability
[9]. However, the conclusions vary considerably, depending on the exact composition of the rock considered. A common finding is a reduction in permeability, which has been attributed to mechanisms such as the precipitation of secondary minerals
[7],
[8],
[10],
[11], the dislodging and blockage of clay and quartz particles
[12],
[13],
[14], mechanical weakening and compaction
[5], and hydrodynamic effects
[15]. Dávila et al.
[16] reported decreases in both porosity and permeability, with simulations suggesting the precipitation of montmorillonite and mesolite, alongside the dissolution of K-feldspar, calcite, and illite. In contrast, Lamy-Chappuis et al.
[17], Zou et al.
[18], Tang et al.
[19], Sun et al.
[20], and Gholami and Raza
[21] reported increased permeability following CO
2 injection, attributing it to the dissolution of carbonate cement, alkali feldspar, and clay minerals. Sayegh et al.
[22] noted an initial permeability reduction due to fines migration blocking the pore throats, followed by an increase in permeability as these fines were later dissolved. Al-Yaseri et al.
[23] suggested that permeability is influenced by both sandstone mineralogy and brine composition, with permeability increasing during low-salinity CO
2-saturated water injection but decreasing under high-salinity conditions.
Less work has explicitly studied the effect of reactions on multiphase flow properties. Ge et al.
[24] proposed that CO
2-saturated water injection induces fines migration and mineral reactions, resulting in a reduction in CO
2 drainage relative permeability, while the relative permeability of the water remains unaffected. In a related study, Gholami and Raza
[21] observed that calcite precipitation, clay dissolution, and a reduction in quartz hydrophilicity collectively increased porosity and absolute permeability but decreased CO
2 relative permeability. Similarly, Kou et al.
[25] found that carbonate mineral dissolution dominated over precipitation, leading to increased porosity and permeability, enhanced CO
2 wettability, reduced brine relative permeability, and increased CO
2 relative permeability. Sun et al.
[20] suggested that the reaction between carbonic acid and tight sandstone during CO
2 injection increased porosity and permeability, reduced CO
2 relative permeability, and initially decreased brine relative permeability at low CO
2 saturation, which then increased as CO
2 saturation rises. These studies highlight the variability and inconclusiveness of laboratory results, underscoring the need for further investigation into the precise characteristics of reactive flow and its underlying mechanisms. In particular, the effect of mineralogical changes on relative permeability in a reservoir sandstone—representative of a putative storage formation—has not been investigated and is thus the focus of this study.
The exact reactions among CO
2, brine, and sandstone have been extensively studied in autoclaves and batch reactors. Liu et al.
[26] injected CO
2 into sandstone/hot water systems and observed sandstone dissolution, along with the deposition of aluminum silicate and calcium–aluminosilicate secondary minerals. Kaszuba et al.
[27] and Ilgen et al.
[28] found magnesite precipitation and noted the dissolution of silicate minerals and calcite/dolomite, respectively, in CO
2 treatment. Fu et al.
[29] and Lu et al.
[30] reported extensive feldspar dissolution and secondary mineral precipitation involving iron minerals and kaolinite on feldspar surfaces after CO
2 exposure. Furthermore, Dawson et al.
[31] and Pearce et al.
[32],
[33] found that, while carbonate cements dissolved, low-surface-area K-feldspar grains remained unreactive in Berea sandstone exposed to CO
2 in sodium chloride (NaCl) brine. They also observed the dissolution and precipitation of carbonate and silicate minerals, along with the precipitation of barite, iron (Fe)-oxides, clays, and gypsum, as well as ion leaching from Fe-rich chlorite, clay structural collapse, and fines migration during impure CO
2 injection into clay-rich mudstone cores. Carroll et al.
[34] proposed that clay dissolution and secondary mineral precipitation, including amorphous silica and kaolinite, were dominant during CO
2–mineral reactions. Fuchs et al.
[35] aged sandstone in CO
2–saturated brine for four or eight weeks and revealed the loss of clay cementation, greater exposure of quartz and K-feldspar grains, and apparent surface roughening. While these studies offer valuable insights into the mineral reactions occurring during CO
2 injection, they fail to quantitatively correlate these reactions with the resulting alterations in petrophysical properties, and they do not sufficiently capture the complexities of multiphase reactive flow in sandstone formations.
In light of these findings, we investigate the characteristics and mechanisms of multiphase reactive flow during CO2 storage in sandstone through steady-state relative permeability experiments, in situ imaging, and ex situ scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS) analyses. Our objectives are threefold: ① to characterize multiphase reactive flow behavior, ② to identify alterations in pore structure, and ③ to unravel the underlying CO2–brine–rock reactions during CO2 injection into sandstone.
2. Materials and methods
2.1. Rock samples and fluid properties
We used a cylindrical sandstone sample sourced from a depleted gas reservoir, with a diameter of (6.03 ± 0.05) mm and a length of (20.05 ± 0.05) mm. The sample exhibited a helium porosity of 25.0%, and its absolute permeability was measured to be (185 ± 10) mD; 1 mD = 0.987 × 10−11 cm2.
The brine used in this experiment was deionized water doped with 30 wt% potassium iodide (KI). This concentration was selected to facilitate phase segmentation in image analysis, following the methodology of Lin et al.
[36]. At the experimental conditions of 50 °C and 8 MPa, the viscosities of the CO
2 and brine were (0.020 ± 0.001)
[37] and (0.60 ± 0.05) mPa·s
[38], respectively, with an interfacial tension of approximately 35 mN·m
−1 [39].
2.2. Experimental methods
2.2.1. Steady-state relative permeability experiment
The sample was dried for 24 h in a vacuum oven, enclosed in a Viton sleeve capped with end fittings connected to tubing, and assembled within an aluminum core holder. The core holder was placed inside a micro-computed tomography (micro-CT) scanner (Zeiss Xradia Versa 510), with tubing connections established to the brine and CO
2 supplies, the receiving and confining pumps, and the reactor and pressure transducer. A water bath circulator was used to maintain the temperature in the pumps, while a heating jacket, controlled by a proportional–integral–derivative (PID) system
[40],
[41], regulated the temperature of the core holder, as shown in
Fig. 1. To minimize mass transfer between the CO
2 and brine during the experiment, the brine was pre-saturated with CO
2, and the CO
2 was pre-saturated with brine inside the reactor.
A confining pressure of 2.0 MPa was applied, and the sample was initially scanned at a voxel size of 6.5 µm for saturation analysis. Subsequently, a higher-resolution scan (voxel size: 2.9 µm) was performed to assess pore structure, contact angle, and curvature. CO
2 was injected for 30 min to dislodge any fine particles within the sample, followed by a 12 h system vacuuming. Brine was then injected at flow rates of 0.20, 0.50, and 1.00 mL·min
−1, while the pressure differential was monitored to determine the absolute permeability using Darcy’s law. Afterward, the injection rate was reduced to 0.20 mL·min
−1, and scanning was conducted to capture a fully brine-saturated image. Relative permeability was measured using the steady-state method across a series of brine fractional flows (the volume fraction of brine injected,
fw = 0, 0.05, 0.10, 0.30, 0.70, 0.90, and 1.00), maintaining a constant total flow rate of 0.20 mL·min
−1. The data at
fw = 1
.00 was obtained from another core drilled adjacently, with similar porosity and permeability, and subjected to the same experimental procedure. Each fractional flow required approximately 24 h to reach a steady state, after which the sample was scanned using the same imaging protocol. The experimental procedures were consistent with those reported by Krevor et al.
[42] and Gao et al.
[43].
After completing all the fractional flow experiments, the sample was extracted, and the pressure drop along the tubing was measured. These pressure drops were subtracted from the measured pressure differences to calculate the absolute and relative permeability. The extracted sample was then dried in a vacuum oven for 24 h, after which a dry scan was conducted to assess changes in pore structure and mineralogy. Finally, the sample’s absolute permeability was measured as previously described.
For imaging, we used a Zeiss XRM-510 micro-CT scanner with an X-ray energy of 80.0 keV and a power setting of 7.0 W. Each scan comprised 3201 projections, with an exposure time of 2 s and a binning factor of 2. Image reconstruction was carried out using Zeiss Reconstructor Software after correction of the center shift and beam hardening. To generate a complete image of the sample, four sectional images were normalized and stitched together, resulting in dimensions of 1025 × 1054 × 2751 voxels. A series of high-resolution, zoomed-in images with dimensions of 500 × 500 × 500 voxels were also obtained. To ensure consistent orientation throughout the experiment, all images were registered to the dry sample scans, with the Lanczos algorithm being employed for image resampling. Noise reduction and image smoothing were achieved using a non-local means filter, which preserved edge details.
Image segmentation was performed using a combination of differential imaging, interactive thresholding, and interactive top-hat filtering to distinguish the three phases: rock, brine, and CO
2. For each fractional flow, the brine phase was isolated by subtracting the dry image from the raw image, whereas the CO
2 phase was isolated by subtracting the raw image from the brine-saturated image. Large ganglia of the brine or CO
2 phases were identified through interactive thresholding, while small ganglia were captured using top-hat filtering. The segmented large and small ganglia were then merged to reconstruct the brine and CO
2 phases. Saturation was quantified from the segmented core scanning images, while other properties—including the distribution of the pore and throat radii, coordination number, interfacial curvature, contact angle, and shape factor (a measure of pore angularity)—were derived from the segmented high-resolution zoomed-in images, following the methods described by Raeini et al.
[44], AlRatrout et al.
[45], and Foroughi et al.
[46].
2.2.2. SEM and EDS analysis
A Hitachi TM4000Plus SEM was employed to characterize the nanoscale mineral features before and after CO
2 injection. The SEM operated at an acceleration voltage of 15 kV, a beam current of 10 nA, and a temperature of 20 °C. Three adjacent samples, drilled from the same core, were used to observe the common mineral properties prior to CO
2 injection. The sample used in the steady-state imbibition relative permeability experiment was analyzed to assess the mineral characteristics following CO
2 injection. Initially, low-magnification imaging (×30) and EDS mapping were combined to obtain a cross-sectional view of the sample and identify the minerals types and their distribution
[47]. High-magnification imaging was subsequently carried out field by field, combined with point EDS at the center of the minerals, to accurately characterize the diverse mineral features
[48]. To ensure consistency and reliability, these mineral characteristics were consistently observed across multiple samples.
3. Results and discussion
3.1. Reactive flow characteristics
The measured relative permeabilities, along with the corresponding pressure differences and brine saturations, are presented in
Fig. 2 [20],
[21],
[24],
[31],
[49],
[50],
[51]. As shown in
Figs. 2(a) and
(b), at the onset of imbibition (
fw = 0), the irreducible brine saturation is approximately 0.12, and the relative permeability of CO
2 reaches 0.47. This low initial brine saturation suggests a high capacity for CO
2 storage
[49]. When the fractional flow is increased to
fw = 0.05, the brine saturation rises sharply to 0.51, while the CO
2 relative permeability decreases significantly to 0.013. This remarkable increase in brine saturation, coupled with the steep decline in CO
2 relative permeability, is likely due to brine—as the wetting phase—occupying a large fraction of the smaller pores and throats, trapping CO
2 and hindering its flow. As the brine saturation continues to increase, the relative permeability of the brine rises steadily, whereas the relative permeability of the CO
2 declines gradually. During this phase, the brine becomes more connected, facilitating its flow
[49],
[52], while the flow of the CO
2 is progressively impeded, as demonstrated in
Fig. 2(e).
The overall trend of our results is consistent with the special core analysis (SCAL) data from a larger sample (34 mm in diameter) from the same formation in a steady-state drainage experiment, confirming the reproducibility and accuracy of our results. Furthermore, when compared with other studies on sandstones without geochemical reactions
[24],
[42],
[50],
[51], as illustrated in
Fig. 2(a), a key distinguishing feature is the significantly lower CO
2 relative permeability observed under reactive conditions. For instance, in the Tuscaloosa sandstone examined by Krevor et al.
[42], despite comparable petrophysical properties (absolute permeability: 220 vs 185 mD; porosity: 23.6% vs 25.0%), the relative permeability of CO
2 in reactive systems is reduced to approximately one-tenth of that in non-reactive cases. This observation correlates with the findings of Sun et al.
[20] and Ge et al.
[24], whose steady-state and unsteady-state relative permeability experiments on tight sandstone and Berea sandstone, respectively, revealed that reactive flow–rock interactions can increase brine flow capacity and enhance CO
2 trapping. However, the underlying mechanisms may differ, as we observed a decrease in absolute permeability from (185 ± 10) to (110 ± 8) mD during CO
2 treatment, whereas those studies suggested an increase in permeability.
The reduction in CO2 relative permeability, coupled with the decline in absolute permeability, is expected to impair CO2 injectivity and diminish CO2 storage capacity to a certain extent in both laboratory experiments and field-scale operations. A more comprehensive investigation into the petrophysical properties is essential to fully elucidate the mechanisms underlying this phenomenon, as detailed in the following section.
3.2. Analysis of changes in pore structure
Fig. 3 shows images of the sample and an analysis of the pore and throat structures before and after CO
2 injection, revealing significant variations.
Fig. 3(a) highlights pore formation, enlargement, and shrinkage during CO
2 injection, as quantified in
Fig. 3(b). The pore formation and enlargement, reflected in the increased porosity, are likely driven by mineral dissolution and the migration of fines
[20],
[21],
[24]. In contrast, significant pore shrinkage, corresponding to a decrease in porosity, was also observed after CO
2 injection, likely due to secondary mineral precipitation and pore blockage by fines
[25]. While these mineral reaction phenomena have been documented in previous studies using SEM
[29],
[30],
[31],
[32], this study provides the first
in situ pore-scale evidence. Although the average porosity increased, the enhanced heterogeneity impeded multiphase flow, as evidenced by the reduction in permeability.
Pore and throat size, coordination number, and shape factor provide quantitative insights into the alteration of the pore space during CO
2 injection. The coordination number, defined as the number of throats connected to a pore, decreases markedly after CO
2 injection (
Fig. 3(d)), highlighting the reduction in pore connectivity. As shown in
Fig. 3(e), CO
2 injection leads to a slight reduction in both pore and throat sizes, accompanied by a noticeable increase in the number of isolated elements. This indicates a significant reduction in flow pathways during CO
2 injection, consistent with the findings of Meng et al.
[53], who investigated CO
2 storage in nanopores using molecular simulations and convincingly demonstrated, with robust evidence, that smaller pores are more effective for CO
2 trapping. Additionally, the shape factors of both pores and throats
[44], as shown in
Fig. 3(f), decrease, suggesting that the pores become more irregular in shape post-injection. This implies an increase in flow resistance. Consequently, the combined effects of reduced pore and throat sizes, diminished connectivity, and increased irregularity restrict the available flow pathways, leading to elevated flow resistance. This finding clarifies why, despite the increase in average porosity, the absolute permeability decreases.
Previous studies on changes in petrophysical properties have predominantly focused on porosity
[14],
[17],
[18],
[21], with some providing pore size changes using nuclear magnetic resonance (NMR)
[20],
[25]. However, looking at porosity alone is misleading because, even with significant dissolution, the pore becomes less well connected, resulting in decreases in the absolute and relative permeability.
Fig. 4 shows the measured distribution of the contact angle, the angle measured through the brine at which the two fluid phases meet the solid surface
[45],
[54],
[55], and the curvature of the interfaces between the brine and CO
2. Two key trends emerge from
Figs. 4(a) and
(c). First, as the fractional flow of the brine (
fw) increases, the contact angle also increases, with the most frequent contact angle rising from 62.5° at
fw = 0.05 to 71.5° at
fw = 0.90. This suggests that the system is weakly water-wet, and that geochemical reactions in CO
2 injection reduce its water wettability. This observation aligns with the findings of Gholami and Raza
[21], who measured brine droplet contact angles on sandstone before and after CO
2 treatment using the
ex situ drop shape method under ambient conditions. Second, the range of contact angles broadens with increasing
fw, suggesting enhanced wettability heterogeneity, particularly in the less water-wet regions. These changes in wettability are expected to significantly facilitate brine flow and enhance CO
2 trapping, as reflected in the relative permeability measurements. Such alterations are likely driven by complex mineral reactions, which will be discussed in the following section.
As the fractional flow of brine (
fw) increases, the mean curvature of the CO
2–brine interface (
Figs. 4(b) and
(d)) gradually decreases, with the most frequent value declining from 0.046 μm
−1 at
fw = 0.05 to 0.029 μm
−1 at
fw = 0.90. This trend suggests that the interface becomes progressively flatter and smoother, which aligns with the observed reduction in water wettability. Using the Young–Laplace equation, this can be related to a reduction in capillary pressure from 3.2 kPa at
fw = 0.05 to 2.0 kPa at
fw = 0.9, which is as expected showing a decrease in capillary pressure, with saturation as displacement proceeds. Compared with that at a low
fw, CO
2 trapping at a higher
fw with lower capillary pressure is significantly reduced
[25], facilitating easier flow and aligning with the contact angle and relative permeability results.
This paper presents, for the first time, in situ pore-scale characteristics of interface properties, including contact angle and curvature. These findings bridge geophysical property alterations and changes in relative permeability, providing direct pore-scale evidence of the mechanisms driving multiphase flow in subsurface environments.
3.3. Mineral reactions
The mineral compositions of the sandstones, as revealed by SEM and EDS mapping in
Fig. 5, illustrate the mineralogical changes induced by CO
2 injection. Prior to CO
2 exposure, the sandstone is composed of elements including O, Si, C, Fe, Al, Ca, K, Na, and S, as shown in
Fig. 5(a). Based on elemental distributions and previous studies
[47], the dominant minerals are inferred to include quartz, feldspar, albite, kaolinite, calcite, gypsum, and siderite. The presence of carbon is likely due to residual organic matter from the depleted gas reservoir
[56]. Quartz and feldspar form the primary framework minerals, while albite, kaolinite, calcite, and gypsum occur in smaller, more scattered concentrations as cementing phases.
Post-CO
2 injection,
Fig. 5(b) reveals that only O, Si, C, Al, K, and Fe are detected, indicating the presence of quartz, feldspar, kaolinite, and siderite. Notably, Na, Ca, and S—associated with albite, calcite, and gypsum—are absent, suggesting that these minerals have undergone significant reactions and interactions during CO
2 injection. These are the underlying mechanisms driving the aforementioned alterations in petrophysical and interfacial properties, as well as relative permeability
[33],
[48]. The dissolution of calcite during CO
2 injection has been widely reported
[18],
[21],
[25]. The reactions and interactions of albite, gypsum, and other minerals during CO
2 injection will be explored in detail in the following high-resolution images.
As illustrated in
Fig. 6, different minerals exhibit distinct reactions to CO
2 injection. Before CO
2 exposure, quartz presents a smooth, intact surface (
Fig. 6(a)); following exposure, however, minor etch pits appear (
Fig. 6(b)). Feldspar, which shows some surface alteration prior to CO
2 contact while retaining its structural integrity (
Fig. 6(c))
[5],
[51], undergoes more pronounced changes post-injection, including the formation of laminar channels, etch pits, and crystal fragmentation (
Fig. 6(d)). These alterations, coupled with the potential migration of feldspar fragments, likely have a significant impact on pore structure and permeability.
Kaolinite, the primary cementing material, forms a booklet-like structure and extensively fills pores before CO
2 exposure (
Fig. 6(e)). After CO
2 treatment, kaolinite migrates and accumulates, blocking small pores and throats, which increases reservoir heterogeneity and consequently hinders multiphase flow (
Fig. 6(f)), as also observed by Othman et al.
[13] and De Silva et al.
[14] using SEM on post-CO
2 treated sandstone. Moreover, we observed newly formed, small kaolinite crystals “growing” on feldspar, as shown in
Figs. 6(j) and (k). This is likely a result of feldspar and albite reacting with CO
2, leading to their transformation into kaolinite, which contributes to pore blockage and reduced permeability. This reaction plays a critical role in CO
2 storage, in line with the findings of Wang et al.
[57], who were the first to quantify the contributions of mineral precipitation over the entire life cycle of a low-permeability tight sandstone reservoir by innovatively integrating experiments, transient well testing, and numerical simulations. Albite, which initially displays minor alterations as platy crystals with cleavage planes (
Fig. 6(g)), disappears after CO
2 exposure, likely due to its transformation into feldspar through interaction with excess K
+ ions in the experimental brine.
Calcite, which is characterized by its rhombohedral cleavage, acts as a cement binding mineral grains (
Fig. 6(h)). Its dissolution during CO
2 contact is well-documented and may promote fines migration, as suggested by Tang et al.
[12]. Gypsum, which serves a role similar to that of calcite, exhibits some fragmentation even before CO
2 exposure (
Fig. 6(i)), and its disappearance after treatment is likely due to fragmentation and migration caused by prolonged flushing.
4. Conclusions
Multiphase reactive flow plays a critical role in CO2 injection into multi-mineral sandstones, directly affecting injectivity, storage capacity, and the long-term safety of CO2 sequestration. To elucidate its mechanisms and impact, we performed steady-state imbibition relative permeability experiments, integrated with in situ X-ray imaging and ex situ SEM–EDS scanning on a sandstone with complex mineralogy from the putative storage site. The primary findings were as follows.
(1) A pronounced decline was observed in both CO2 relative permeability and absolute permeability during multiphase reactive flow.
(2) Changes in petrophysical properties were identified, including reduced pore and throat sizes, diminished connectivity, and increased pore irregularity.
(3) The primary mechanisms included mineral dissolution, precipitation resulting from feldspar-to-kaolinite transformation, and fines migration.
Our study provides the first comprehensive analysis of multiphase reactive flow, establishing a foundation for improving injectivity, optimizing storage capacity, and ensuring long-term safety, particularly in mineralogically complex sandstone formations. For field-scale operations, it is essential to thoroughly characterize the mineralogical composition. In the case of complex sandstone formations, customized strategies—such as ion-mediated brine injection—should be employed to mitigate adverse reactions, as discussed earlier. Future research should focus on developing quantitative models to elucidate the relationships between permeability, petrophysical properties, and geochemical reactions, thereby improving predictions of CO2 storage efficiency and safety in heterogeneous reservoirs.
CRediT authorship contribution statement
Rukuan Chai: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Qianqian Ma: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Sepideh Goodarzi: Methodology, Investigation, Formal analysis. Foo Yoong Yow: Writing – review & editing, Resources, Methodology. Branko Bijeljic: Writing – review & editing, Validation, Supervision, Methodology, Conceptualization. Martin J. Blunt: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Conceptualization.
Declaration of competing 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.
Acknowledgments
The authors gratefully acknowledge PETRONAS for their financial support and permission to publish this paper. We also extend our sincere gratitude to Anfal Al Zarafi, Anindityo Patmonoaji, Shanlin Ye, Edward Bailey, and Vincenzo Cunsolo for their invaluable assistance in conducting the experiments and analyzing the results.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2025.01.016.
PDF
(5185KB)
