Fractures formed through supercritical CO₂ fracturing exhibit distinct characteristics that are influenced by the properties of supercritical CO₂. One notable effect is the reduction in the breakdown pressure of the formation, allowing for easier penetration into the rock matrix due to its low viscosity, high diffusivity, and absence of surface tension. Supercritical CO₂ exhibits a lower viscosity (three magnitudes or even smaller) compared to alternative fracturing fluids like water-based fluids (Kuang et al., 2023). This attribute allows for smoother flow through minuscule pores and fractures within the reservoir rock, enabling deeper penetration into the rock matrix and generating fractures with enhanced tortuosity. This characteristic allows the injected fluid to effectively access the pre-existing fracture, and as the induced fracture propagates along its path, there is a significant reduction (∼50%) in breakdown pressure (Zhang et al., 2017a; Li et al., 2019; Zhang et al., 2021; Feng and Firoozabadi, 2023), as summarized in Table 2.

Laboratory tests have indicated that fractures created by CO₂ have higher tortuosity, as illustrated in Figure 2 (Song et al., 2019). Tortuosity pertains to the extent of deviation from a linear trajectory observed in fractures. The higher degree of fracture tortuosity indicates that CO₂ follow intricate routes within the reservoir rock, thereby augmenting their interaction surface. Additionally, CO₂ fracturing holds the potential to generate a more intricate network of interconnected fractures within the reservoir, surpassing the complexity of fractures induced by water-based fluids (Wang and Sharma, 2023). This characteristic enhances the fracture surface area and the volume of the reservoir that is stimulated. The increased fracture surface area provides more flowing pathways for in situ hydrocarbons during production operations, thus improving well productivity and enhanced recovery rates from unconventional formations.

However, a significant challenge associated with CO₂-created fractures is their underdeveloped width, primarily due to the high rate of CO₂ leak-off and the net stress loss within the fracture. This loss of driving force leads to narrower and shorter fractures (Zhou and Burbey, 2014; Wang et al., 2017; Ranjith et al., 2019). The average fracture aperture of water, N₂, L-CO₂ and SC-CO₂ shows relatively small variances falling in the range between 0.304 mm and 0.317 mm, as presented in Figure 3. However, the largest standard deviation (0.201) of the aperture formed by SC-CO₂ fracturing is obtained, followed by water fracturing (0.171), L-CO₂ fracturing (0.123), and N₂ fracturing (0.091). This suggests the maximum roughness of the fractures created by SC-CO₂ (Yang et al., 2021). Furthermore, given the higher complexity and tortuosity of CO₂-created fractures, the injection of proppants afterward becomes more challenging, resulting in elevated operation wellhead pressures. More careful planning and innovative solutions tailored specifically for CO₂ fracturing techniques are essential to overcome these challenges posed by narrow and short CO₂-created fractures with complex geometries.

In addition to the more challenging conditions for proppant transport, a significant hurdle in CO₂ fracturing is the low viscosity of supercritical CO₂, which is similar to gaseous CO₂. As a result, proppant particles settle rapidly, leading to the formation of accumulations known as dunes (Hou et al., 2015). These dunes vary in shape and size as continuous injections progress. Within fractures, the proppant is then transported in the form of these dunes, creating a dynamic and complex process, unlike water-based high-viscosity fluids that evenly suspend the proppant (Hou et al., 2022a; Hou et al., 2022b). When the mass flow remains constant, altering the injection temperature to a higher value or reducing the injection pressure will lead to a decrease in both viscosity and density of supercritical CO₂, resulting in evolutions of equilibrium height and distance for dune transport, as presented in Figure 4 (Zheng et al., 2022b).

Furthermore, the high leak-off of CO₂ and the distribution of the injected fluid in complex fracture networks exacerbate proppant accumulation and can even cause sand screen-out due to the loss of the carrying fluid. Previous studies have demonstrated the influence of supercritical CO₂ on the settling, restarting, and flowing behaviors of proppants (Hou et al., 2017a; Chen and Sun, 2023). It has been observed that the high density of CO₂, which is similar to liquid CO₂, contributes to an enhanced capacity for proppant transport in supercritical CO₂, as depicted in Figure 5. Each black point (P1, P2, P3, P4 and P5) represents a proppant particle captured by the high-speed camera. The dashed lines derived from the black points represent the moving trajectory of the proppant particles, which are plotted automatically by the image analysis software. The terminal settling velocity of proppants in CO₂ is slightly higher, within the same magnitude, compared to settling velocities in water (Hou et al., 2015). Additionally, the slippage between the particles and the carrying CO₂ can be eliminated by increasing the flow rate of the slurry (Hou et al., 2017b). Restarting the movement of particles in CO₂ is even easier than in water due to the absence of interfacial tension and the generation of additional Magnus force through high-speed spinning, facilitating the restarting process (Hou et al., 2019).

The primary objectives of a fracturing fluid are to create fractures and transport proppants. However, CO₂ fracturing faces significant challenges in both areas, as outlined in Figure 6. In terms of fracture creation, there are several approaches that can be employed to improve performance. These include reducing the leakage of CO₂ into the rock matrix and natural fractures, establishing net stress within the fractures, and then enhancing the propagation of fracture networks. On the other hand, the capacity of CO₂ to carry proppants can be enhanced by addressing issues such as particle settling, eliminating slippage between the particles and CO₂, and improving proppant transport within complex fracture networks.

In order to address these challenges, various solutions have been analyzed and summarized in Figure 6. One common approach is the use of CO₂ thickeners, which increase the viscosity of the fluid. This serves to reduce both proppant settling and CO₂ leak-off (Enick et al., 2012; Al Hinai et al., 2018). Additionally, it has been observed through numerical and experimental simulations that fracture width and particle slippage are influenced by the CO₂ pump rate. Higher pump rates facilitate fracture growth and help eliminate slippage (Lei et al., 2016). CO₂ leak-off is another prevalent issue that hampers fracture propagation and proppant transport. Therefore, promising research directions for improving CO₂ as a fracturing fluid include the development of friction reducers, investigating the effects of flow loss (caused by leak-off) and distribution (within fracture networks), as well as studying the shear viscosity of thickened CO₂ (the performance of the thickener under high pump-rate condition).

The CO₂ fracturing technique is mainly applied in unconventional formations that have extremely tight rock matrix and nano-Darcy permeability (decreasing the leak-off of CO₂). Three representative cases of CO₂ fracturing in tight oil, shale gas, and shale oil formations are summarized in Table 3 (Meng et al., 2016; Yiyu et al., 2021; Jing et al., 2022). Case A uses pure CO₂ for cracking fractures and carrying proppant. Two types of additives are tested to increase the viscosity of CO₂ and its proppant-carrying capacity. Cases B and C only use CO₂ to crack the formation and create complex fracture networks. The high-viscosity gel is applied afterward to further develop the networks and carry the proppant. Generally, both the fracturing scale and pump rate are relatively small for CO₂ fracturing compared with those for water-based fracturing.

Approximately ∼300 m³ of CO₂ is injected in each well or stage, with approximately ∼180 m³ of CO₂ injected in each stage of the horizontal well in Case C. The pump rate may be restricted by the high friction encountered along the wellbore when using CO₂, resulting in a high wellhead pressure of approximately ∼65 MPa, as indicated in Table 3. The sand ratio in Case A is around 5.6%, which is less than half of the sand ratio typically used in water-based fracturing. The efficiency of CO₂ fracturing in field trials is relatively low due to the limited scale of fracturing (both proppant and CO₂ volumes), low sand ratio, restricted pump rate, and comparatively high injection pressure. This could be one of the main reasons, as well as the high cost of CO₂ additives, why recent tests have opted for a hybrid approach that combines CO₂ injection with water-based fracturing, as illustrated in Cases B and C.

Most of the current CO₂ fracturing field cases are reported along with the development of shale oil in China (Hou et al., 2024a). The field engineers injected CO₂ as a pre-fracturing process, aiming to create more complex fracture networks. The following injected conventional water-based fluids continuously develop the fracture dimensions and transport the proppant into fractures (Yang et al., 2022). Therefore, the usage of CO₂ (several hundred scales for each fracturing stage) is significantly smaller than the water-based fluids for the main fracturing operation, as listed in Table 3. However, the flowback rate of fracturing injected CO₂ is approximately one order of magnitude lower than other forms of CO₂ storage (as shown in Figure 7), indicating a higher efficiency of permanent CO₂ storage (Louk et al., 2017; Hou et al., 2024a). The usage of CO₂ may be improved by increasing its proportion in the total fracturing fluids. A possible approach is using CO₂ to share the proppant injection task, for instance, carrying the fine proppant (100 mesh) (Hou et al., 2017a; Hou et al., 2017b). Meanwhile, the developments of the carbon market and carbon capture techniques may reduce the cost of CO₂ sources. The policy incentives are also essential to encourage the operators to promote the usage of CO₂, for instance, the tax preference applied in the United States (Ren et al., 2022). The increasing proportion and decreasing cost may significantly promote the potential and contribution of CO₂ fracturing to CCUS, considering the huge consumption of fracturing fluids.

The other challenge of CO₂ fracturing for CCUS is the mismatch between CO₂ sources and fracturing sites (Munkejord et al., 2016), for instance, transporting the captured CO₂ from power plants to oil and gas fields. Pipelines may be necessary for the continuous transport of CO₂ for huff-and-puff, EOR or direct storage in relatively fixed sites (Onyebuchi et al., 2018). Trucks may be essential for fracturing operations to transport CO₂ from one site to another (Gao et al., 2011). Both pipelines and trucks will increase the investments in construction and equipment, as well as the potential for extra CO₂ emissions. Therefore, CO₂ transport has become a common issue for all kinds of CO₂ storage because of the geographical distance between CO₂ sources and storage sites. For CO₂ fracturing, a hybrid transport system may be a solution to improve the flexibility of CO₂ transport from site to site.

Although fundamental research has highlighted the advantages and feasibility of CO₂ fracturing, field trials have encountered significant challenges, as summarized in Table 3. To reveal the disparity between the theory and practice of CO₂ fracturing, we compared key injection parameters recorded during field operations using different fracturing fluids, as presented in Figure 8. Initially, in conventional reservoirs, guar gel (referred to as the first generation of fracturing fluid) was used to create large bi-wing fractures with a high concentration of large proppants. Subsequently, slickwater (with lower viscosity) was employed at a much higher pump rate to carry smaller proppants at lower concentrations, achieving a balance between fracturing efficiency and investment-production ratio (Barati and Liang, 2014; Zhang et al., 2017b). CO₂ fracturing, known as the third generation of fracturing fluid, is considered environmentally friendly. Pump rates and injection scales are both reduced for CO₂ fracturing, reflecting its status as a developing technique.

Compared to conventional fluids, the disparity between fundamental research and field application primarily lies in pump rate and operation scale, as depicted in Figure 8. Current efforts to thicken CO₂ may draw inspiration from the success of first-generation fluids, characterized by high viscosity gel. However, the second-generation fluid (represented by low-viscosity slickwater) compensates for the low-viscosity drawback with a high pump rate, which offers valuable insights. Therefore, the utilization of friction reducers becomes another crucial technique for CO₂ fracturing. Correspondingly, the performance of CO₂ thickener (enhanced CO₂ viscosity after the high-pump-rate shear) becomes an essential criterion for the relevant research, which currently is barely reported. Other valuable insights include enhancing fracturing scales through the development of low-cost additives, increasing the proportion of fine proppant, and adopting hybrid approaches that incorporate water-based fluids (inspired by Cases B and C).

The relatively low pump rate may represent one of the most significant gaps between the theory and practice of CO₂ fracturing. Firstly, the proppant usually settles down rapidly in low-viscosity fluids (CO₂ and slickwater). The horizontal transport distance of the proppant before its settlement reduces under a low pump rate condition due to the lower horizontal dragging force (Hou et al., 2017b; Hou et al., 2019). This significantly constrains the proppant transport capability of supercritical CO₂, and then the scale of proppant injection in fields (Table 3). Secondly, the high diffusion feature of supercritical CO₂ induces a high leak-off of fluid from fractures into the formation. The low pump rate may weaken the supplementary fluid in fractures, thus constraining the propagation of fracture networks. Meanwhile, the relatively low fracturing scale further deteriorates the development of underground fractures. Correspondingly, the stimulated reservoir volume is restricted for enhancing oil/gas production. Regarding the CO₂ storage concern, the low fracturing scale reduces the usage of CO₂ during fracturing operations. The limited artificial fracture volume will further decrease the inventory capacity of CO₂ storage in unconventional reservoirs, because the artificial fracture may contribute most to the capacity of CO₂ storage (Hou et al., 2024a; Hou et al., 2024b). Therefore, the relatively low fracturing scale may represent the other critical gap between the theory and practice of CO₂ fracturing in accordance with the aforementioned rationale.