The study area is located in the eastern Ordos Basin (Figures 1A, B). As a part of Ordos Basin, the study area experienced multiple episodes of regression event, accompanied by deposition of marine, transitional, and continental sediments during the Late Carboniferous to the Middle Permian (Chen et al., 2011). The Lower Permian Shanxi Formation, as the target of this study, is in conformable contact with both the underlying Taiyuan Formation and the overlying Shihezi Formation (Figure 1C) (Li et al., 2019). The top of the Taiyuan Formation is suggested to be formed in the open platform sedimentary environment of the shallow shelf facies (Kuang et al., 2020), because it is lithologically dominated by the micritic bioclastic limestone with normal-size fusulinid, echinoderms, and brachiopods as the main bioclastic particles. During the sedimentary period of Early Permian Shanxi Formation, the study area is dominated by marine–continental transitional sedimentary environment consisting of tide-dominated estuarine bay facies and tide-dominated delta facies (Wu et al., 2021). The bottom of Shihezi Formation is mainly medium sandstone, indicating braided river systems (Li et al., 2018; Kuang et al., 2020).

A total of 175 core samples of Shanxi Formation shale were collected from Wells A, B, and C for thin-section identification, X-ray diffraction, and TOC measurement. The whole-rock and clay mineral X-ray diffraction analysis was completed at the PetroChina Research Institute of Petroleum Exploration and Development on a Japan-made physical-science X-ray diffractometer. TOC analysis was performed at the Unconventional Experimental Center of CNOOC Energy Development Co. Ltd using a CS744-MHPC carbon/sulfur analyzer. Ten core-plug samples of different shale lithofacies were implemented with NMR analysis in both dry and saturated fluid states (including n-dodecane and brine) using a c12-010 V low-field NMR device manufactured by the Suzhou Newmai Company. Mercury intrusion was performed using a Quantachrome Poremaster. Samples were prepared to have an approximate size of 20 mm × 20 mm, weighed out to 10–20 g, and then dried at 110°C for at least 24 h under vacuum in an oven. The mercury injection pressure ranged from 0 to 215 MPa in this experiment. The remaining samples were divided into three parts for maceral identification after kerogen extraction, FE-SEM identification, and N₂ and CO₂ adsorption experiments. N₂ and CO₂ adsorption experiments were carried out by crushing the sample into 60–80 mesh, dried in an oven at 110°C for 12 h, and then placed in the Autosorb-IQ3 specific surface and PSD analyzer manufactured by the Cantor Company of the United States. The pretreatment was completed by degassing at 110°C for 12 h in the vacuum condition, and then nitrogen carbon adsorption was carried out. After the experiment, the Brunauer–Emmett–Teller (BET) model was used to calculate the specific SA, and the Barrett–Joyner–Halenda (BJH) model was employed to obtain the PSD and volume. The pore size was divided into three categories according to the pore size classification scheme (Loucks et al., 2012), namely, micropores (<2 nm), mesopores (2–50 nm), and macropores (>50 nm). Hysteresis loops were observed at the relative pressure (P/P_O) of approximately 0.5 in adsorption–desorption curves, which indicates the presence of great differences in pore size and pore morphology among shale samples that cause such adsorption behaviors at this pressure. This relative pressure was set as the threshold to divide the transitional facies shale into two groups. The first group with P/P_O = 0–0.5 is subjected to the monolayer–multilayer adsorption process controlled by van der Waals force, while the second group with P/P_O = 0.5–1.0 experiences the capillary condensation adsorption process controlled by surface tension (Wang M. et al., 2015; Sun et al., 2015). In this paper, the Frenkel–Halsey–Hill (FHH) model is used to calculate the fractal dimensions of these two groups of samples separately, which are denoted as D₂ and D₁ for the first and second groups, respectively.

Lithofacies refers to the mineral composition, color, grain size distribution, and other characteristics of sedimentary rocks (Wang and Carr 2012), which is importantly helpful for the reconstruction of depositional processes and environments (Singh et al., 2009; Jiang et al., 2013). The lithofacies is divided according to the following steps: 1) clay >75%, divided into clayey shale (lithofacies I) using contents of clay, carbonate, and siliceous (quartz + feldspar) minerals as three end members; 2) according to RQC (ratio of siliceous mineral/carbonate mineral) (Jiang et al., 2016), the lithofacies with clay mineral content <75% are divided into siliceous clayey shale (lithofacies II), calcareous clayey shale (lithofacies III), siliceous shale (lithofacies IV), calcareous siliceous shale (lithofacies V), siliceous calcareous shale (lithofacies VI), and calcareous shale (lithofacies VII), while the non-shale facies are siliceous rock (lithofacies VIII) and carbonate rock (lithofacies IX). As shown in Figure 2, Shanxi Formation shale of the study area comprises lithofacies I, II, IV, V, and VI. There were no positive correlations between the contents of siliceous minerals and TOC in all samples (Figure 3A) and no negative correlations between the contents of clay minerals and TOC (Figure 3B). Therefore, the main source of siliceous materials in Shanxi Formation shale is not of biogenic origin (Liu et al., 2017; Huo et al., 2020).

The formation of microfractures in the Shanxi Formation transitional facies shale is often related to the late diagenesis of clay minerals and organic matter. Interlaminar microfractures are commonly developed among clay minerals (Figure 5G), which are formed by internal stress caused by water loss, uniform shrinkage, dry cracking, and recrystallization of clay minerals under overlying formation pressure during diagenesis. Their genesis is similar to that of interlaminar pores (Figure 5H). These interlaminar microfractures normally have flat and smooth or slightly curved shapes and good extensibility, with their length of more than 10 μm and width of less than 200 nm (Figure 5G). In addition, a small number of microfractures are produced by hydrocarbon generation in the interior of organic particles, and contraction fractures are developed at the mineral boundaries. In contrast, organic pores are poorly developed (Figure 5I), which can be attributed to the dominance of vitrinite in kerogen (Cao et al., 2018). The width of these microfractures is relatively large, generally in the range of 500–1,000 nm, and the elongation is short, generally several microns (Figure 5I). Microfractures are generally open and can be connected with other types of pores to form complex three-dimensional pore networks, which are conducive to both free gas enrichment and shale gas seepage and migration, thereby playing a key role in the exploitation and development of shale gas.

Development degree of organic pore, inorganic pore, and microfracture.

Organic pores are generally lipophilic, while inorganic pores are mostly hydrophilic (Li et al., 2016). Accordingly, NMR experiments were carried out under water- and oil-saturated conditions, respectively, to observe the signal characteristics on the transverse relaxation time (T₂) distribution spectra of two types of pores. The presence of three peaks in the T₂ spectra of lipophilic pores indicates three types of organic pores, namely, volumetrically dominant small pores with short T₂, large pores with long T₂, and microfracture developed in organic matter (Figure 6). The development degree of the three types of lipophilic pores varies significantly in different shale lithofacies, with a higher degree seen in lithofacies V, lithofacies VI, and lithofacies I.

The main peak at a T₂ value of 0.5 ms in the water-saturated condition represents the hydrophilic pores, while the secondary peak stands for the microfractures with large pore sizes and small volumes (Figure 6). The development degree of microfractures significantly varies from one lithofacies to another (Figure 6). Specifically, the development degree of inorganic microfractures is the lowest in lithofacies IV and the highest in lithofacies II. The Image J software was used to scan and identify different pore types in FE-SEM images, and results show that organic pores are the most abundant in lithofacies V and lithofacies VI, with an average proportion of 32.7%, while their proportions in other lithofacies range from 24.8% to 28.6%. Inorganic pores are the most developed in lithofacies IV, accounting for 63.2% on average, while their proportions in other lithofacies vary in the range of 46.3–57.7% (Table 1).

Compared with marine shale, the transitional facies shale has higher clay mineral contents as well as more diverse and developed inorganic pore types. The clay mineral content presents a slightly positive correlation with the specific SA (R ² = 0.37, Figure 11D), no significant correlation with the total PV (Figure 11E), a positive correlation with D₁ (R ² = 0.73, Figures 11A,F), and a weak positive correlation with D₂ (R ² = 0.42, Figure 11F). Therefore, the clay mineral content has overall significant controls on the pore structure of the Shan23 shale, especially on the development of inorganic pores.