Marine-continental transitional shale is important for oil and gas exploration in China. This kind of shale is widely distributed in North China, South China, Junggar Basin and Tarim Basin^[1], with a resource volume of approximately 19.8×10¹² m³. It is the important target in future shale gas exploration^[2]. Lithofacies involves rock type, lithologic assemblage and sedimentary structure characteristics formed in a particular sedimentary environment, determines the distribution of “sweet-spot” intervals, controls the hydrocarbon generation capacity and reservoir properties, and affects the fracturability of shales^[3]. In recent years, studies of marine shale formations have made substantial progress in identifying lithofacies characteristics and distribution, stratigraphic settings, favorable lithofacies types, and their control on reservoirs^[4,5,6]. However, research on marine-continental transitional shales has just taken its first step. The Permian system on the eastern edge of the Ordos Basin is rich in marine-continental transitional shale resources, and it is therefore worth more studies for exploration and development^{[2, 7]}. The study on lithofacies not only greatly enriches the understanding of shale sedimentology and reservoir geology, but also is the key to improving the scale and efficiency of shale gas exploration and development^{[7, 8]}.

Present studies on marine-continental transitional shales have mostly concentrated on the basic characteristics, distribution and resource potential evaluation, but less on lithofacies classification, characterization and systematic evaluation. Only a couple of studies on lithofacies classification are based on qualitative indicators such as color, texture, and sedimentary structure^[9,10,11]. However, these studies are unable to provide a full characterization of shale reservoirs, as the methods they proposed are not operable and accurate enough, or without considering the total organic carbon (TOC) index. In terms of the research on marine-continental transitional shales, three issues should be solved: (1) the lithofacies division lithofacies types and characteristics; (2) reservoir space types and pore structures of different lithofacies; (3) favorable lithofacies types and genetic mechanisms. This paper examines the lithofacies types, characteristics, favorable lithofacies types, and genetic mechanism of marine-continental transitional shales in the Permian Shanxi Formation in the eastern margin of the Ordos Basin.

The eastern margin of the Ordos Basin is a slender arc about 450 km long with a width of 26-100 km, and it covers an area of approximately 45 000 km^{2 [2]} (Fig. 1a). During the deposition of the Late Carboniferous Benxi Formation, the North China platform continued to subside, while the North China sea transgressed into the eastern part of the basin in N-S direction, creating a littoral shallow marine depositional environment. During the deposition of the Early Permian Taiyuan Formation, the transgression expanded and overlaid the central paleo-uplift to produce a unified epicontinental sea. During the deposition of the Early Permian Shanxi Formation, the Hercynian tectonic movement caused the entire uplift of North China platform. The seawater gradually regressed from the east and west sides of basin, leaving a marine-continental transitional deposition under an epicontinental sea context. Affected by tidal processes, a shallow sea-lagoon-tide-dominated deltaic sedimentary system developed. By the time the Middle Permian Lower Shihezi Formation deposited, the seawater had totally regressed, marking the beginning of the continental freshwater lake basin deposition and evolution period^[12].

During the deposition of the Early Permian Shanxi Formation, the Ordos Basin was transitioning from marine-continental transitional facies to continental facies. Several sets of organic-rich shale were deposited, with a cumulative thickness of 43.5-187.3 m and an average thickness of 88.6 m^[2]. According to the lithology and sedimentary cycles, the Shanxi Formation can be divided into two members (from the bottom up): Shan₂ and Shan₁ (Fig. 1b). In the study area, the Shan₂ shale is the most typical. Its sedimentary centers are around Yulin-Jiaxian, Qingjian-Daning and Jixian-Hejin, with a cumulative shale thickness of up to 50 m (Fig. 1a). Shan₂ can be further divided into three submembers from the bottom to the top: Shan₂³, Shan₂², and Shan₂¹. The shale of Shan₂³ is stably distributed with a thickness ranging from 20 m to 40 m, and contains a few thin interlayers. Shan₂³ is the key target for exploring and developing the marine-continental transitional shale gas in the eastern margin of the Ordos Basin.

The samples were sourced from two typical marine- continental transitional shale gas wells, Daji51 and Daji3-4 in Daning-Jixian area. The final drilling depth of Daji51 is 2685 m in Ordovician Majiagou Formation, and the target formation is Shan₂³ sub-member. The 183 samples were collected at 20 cm interval. For Daji3-4, the final drilling depth is 2277 m in Ordovician Majiagou Formation, and 60 representative samples were taken from Shan₂³.

Samples with eight different sizes were prepared for different experimental purposes: (1) 243 powder samples of 75-150 μm (100-200 mesh) for TOC measurement; (2) 243 powder samples of 75 μm (200 mesh) for X-ray diffraction (XRD) mineral analysis and major/trace element measurement; (3) 69 wet kerogen samples for microscopic kerogen testing; (4) 243 samples of 10 mm×20 mm×0.03 mm for thin sections observation; (5) 26 samples of 10 mm×10 mm×2 mm cut in the direction perpendicular to the bedding for argon-ion polished and carbon plated scanning electron microscopy (SEM); (6) 54 plunger samples of 25 mm diameter and 30-35 mm high for helium porosity measurement; (7) 26 powder samples of 150-180 μm (80-100 mesh) for low-temperature gas adsorption testing; (8) 26 particle samples of 10 mm×10 mm×10 mm for high-pressure mercury intrusion (HPMI).

TOC, XRD and FIB-SEM experiments were performed at National Energy Shale Gas R&D (Experiment) Center, using a LECCO CS230 carbon-sulfur tester from U.S., a TTRIII automatic X-ray diffractometer from Japan, and an FEI Helios NanoLab 650 dual-beam FIB scanning electron microscope. Major/trace elements were measured at Keyuan Engineering Technology Testing Center, Sichuan Province, using an X-ray fluorescence spectrometer and an inductively coupled plasma atomic emission mass spectrometer. Low-temperature gas adsorption and HPMI experiments were performed at Beijing Center for Physical & Chemical Analysis, using an Autosorb-IQ-MP specific surface area and pore size analyzer and a PoreMaster GT60 instrument covering pore sizes of 0.35-200 nm and 0.0036-1000 μm. All the experiments were strictly conducted in accordance with applicable national and industry standards.

The key to selecting appropriate lithofacies markers lies in whether they reflect the sedimentary environment and whether they can help guide practical production. More specifically, lithofacies markers are selected according to the following criteria: (1) they effectively reflect the depositional environment; (2) they make lithofacies division simple and obvious, practically useful; (3) they are quantitatively available and highly operable. The mineral composition is a direct indicator of the developmental environment of the rocks and it is the cause of rock diversity. TOC indicates hydrocarbon enrichment of fine-grained sedimentary rocks and it is an important metric for reservoir classification and evaluation^[3]. Both parameters can be accurately obtained through experiments, and they provide accurate lithofacies division. For this reason, in the study, we selected mineral composition and TOC as criteria for shale lithofacies division and established a comprehensive shale lithofacies division scheme for the Shanxi Formation (Fig. 2a).

The mineral composition of the Shanxi Formation shale is primarily siliceous minerals, clay minerals and carbonate minerals. Mineral composition, as the primary indicator of lithofacies division, can be accurately and quantitatively obtained by bulk rock XRD analysis. The shale lithofacies is divided based on the three-end-member diagram covering siliceous minerals (quartz+feldspar), carbonate minerals, and clay minerals (Fig. 2). The shale lithofacies is siliceous shale (S) when siliceous minerals account for more than 50%, clay shale (C) when clay minerals account for more than 50%, carbonaceous shale (CA) when carbonate minerals account for more than 50%, and mixed shale (M) when all three kinds of minerals account for less than 50% but more than 25%. TOC is the second important indicator for lithofacies division. Using TOC of 5.5% and 8.5% as boundary values, there are high TOC (>8.5%) shale (H), medium-TOC (5.5%≤TOC≤8.5%) shale (M), and low-TOC (<5.5%) shale (L) (Fig. 2a). Gas-bearing capacity is the key measure for evaluating whether a shale gas reservoir is worth developing. For the Lower Silurian Longmaxi Formation shale gas exploration in China, 2.0 m³/t is the lowest gas content of the reservoir commercial development^[8]. According to the statistics used in this study, for 17 wells in the Daning-Jixian area, the gas content of the Shan₂³ shale is 1.38-5.66 m³/t. Using 2.0 m³/t as a standard, the gas content is correlated with the TOC: the gas content greater than 2.0 m³/t corresponds to TOC of approximately 8.5%, and gas content of 1.5 m³/t corresponds to TOC of approximately 5.5%. Therefore, the TOC limits are determined to be 5.5% and 8.5%.

Using all the dividers together, six lithofacies were identified in Shan₂³: low-TOC clay shale (C-L), low-TOC siliceous shale (S-L), medium-TOC siliceous shale (S-M), medium-TOC mixed shale (M-M), high TOC siliceous shale (S-H), and high TOC clay shale (C-H) (Fig. 2b).

3.2.1. Low-TOC clay shale (C-L)

The C-L features gray to dark-gray silty shale intercalated by sand strips and displays a large amount of fossilized plant stems and leaves in the cores (Fig. 3a, 3b). Under a microscope, numerous plant charcoals can be observed, which have been metasomatized by siderite, with wavy laminae and signs of bioturbation (Fig. 4a, 4b). The TOC is 1.0%-2.0%. The macerals are primarily exinites (averaging 52%), vitrinites (averaging 37.5%), and a small fraction of inertinites (averaging 10.5%). The kerogen type index (TI) is between -48.3 and -0.8. The clay mineral accounts for 44%-82% (averaging 60.8%), the siliceous mineral accounts for 12%-50% (averaging 36.7%), and the brittleness index is 18%-56% (averaging 39.2%) (Table 1).

3.2.2. Low-TOC siliceous shale (S-L)

The S-L features gray to dark-gray silt-bearing shale with horizontal beddings (Fig. 3c-3d). It contains fine fossilized plant fragments. Under the microscope, dolomite silt laminae are observed alternately with granular charcoals, with lenticular texture locally. Charcoals are evident and fine, with signs of bioturbation (Fig. 4c-4e). The TOC is typically 1.0%-1.5%, and the macerals are primarily exinites (averaging 57%), vitrinites (averaging 32%), and a small fraction of inertinites (averaging 8%) and sapropelites (averaging 3%). The TI is between -0.8 and -0.5, suggesting type III kerogen. The siliceous mineral content is 40%-58% (averaging 50.9%), the clay mineral content is 30%-53% (averaging 41.5%), and the brittleness index is 47%-70% (averaging 58.5%) (Table 1).

3.2.3. Medium-TOC siliceous shale (S-M)

The S-M features grayish-black siliceous shale (Fig. 3e, 3f), in which silt-bearing dolomite laminae overlap with charcoal laminae to form a laminar texture (Fig. 4f-4h). The TOC is 6%-9% (averaging 7.2%). The macerals are exinites (averaging 68%), vitrinites (averaging 19%), and sapropelites (averaging 10%). The TI is 17.8- 30.0, suggesting type-II₂ kerogen. The siliceous mineral content is 63%-73% (averaging 68.1%), the clay mineral content is 20%-29% (averaging 25.9%), and the brittleness index is 68%-80%.

3.2.4. Medium-TOC mixed shale (M-M)

The M-M shale features grayish-black calcareous shale (Fig. 3g) containing large amounts of shell bioclasts. On the thin section, large amounts of calcareous bioclasts (approximately 30%-40%) and small fractions of dolomite bioclasts can be detected (Fig. 4i-4k). The TOC is 6%-8% (averaging 6.5%). the macerals are primarily exinites (averaging 58%), sapropelites (averaging 21.5%), and vitrinites (averaging 17%). The TI is 30.8-36.3, suggesting type-II₂ kerogen. The carbonate, siliceous, and clay minerals accounts for 30%-45%, 20%-30%, and 30%-40%, respectively. The brittleness index is 60%-70% (Table 1).

3.2.5. High-TOC siliceous shale (S-H)

The S-H features blackish-gray shale (Fig. 3h) containing siliceous organisms, such as foraminifera and sponge spicules, with dolomitic bioclasts (approximately 20%- 30%) and a small fraction of calcareous bioclasts (Fig. 4l-4o). The TOC is 10%-12% (averaging 11%). The macerals are primarily exinites (averaging 60%), sapropelites (averaging 14%), and vitrinites (averaging 19%). The TI is 17.0-27.3, suggesting type-II₂ kerogen. The siliceous mineral content is greater than 50%. The carbonate mineral content is 10%-20%, with calcite and dolomite contributing approximately to half and half. The clay mineral content is less than 35%. The brittleness index is 67%-70% (Table 1).

3.2.6. High-TOC clay shale (C-H)

The C-H features blackish-gray carbonaceous shale (Fig. 3i). The TOC is 15%-25% (averaging 20%). The macerals are primarily exinites (averaging 60%), vitrinites (averaging 30%), and inertinites (averaging 10%). The TI is between -3.0 and -2.5, suggesting type III kerogen. The clay mineral and siliceous contents are 65%-85% and 15%-35%, respectively (Table 1). Super slim laminae were found, with argillaceous strips and bioturbation (Fig. 4p).

The sedimentary environment controls the lithofacies types and their longitudinal distribution^[3]. In a marine- continental transitional environment, where the hydrodynamic conditions change rapidly, the longitudinal evolution of lithofacies is even more complex. Based on the lithological, sedimentary tectonic, geochemical, and paleontological characteristics, it can be determined that the Shanxi Formation emerged in a sedimentary environment in which the littoral shallow marine facies and deltaic facies coexisted. The Shan₂³ sub-member shale formed in a tide-dominated delta-estuarine sedimentary environment^[7]. The tide-dominated deltaic facies includes three subfacies: tide-affected upper delta plain, tide-dominated lower delta plain, and tide-dominated delta front. The tide-dominated estuarine facies includes five subfacies: tidal flat, swamp, barrier island, lagoon, and bay. Fig. 5 shows the shale characteristics and development locations.

The Sr/Ba element ratio provides effective description of paleo-salinity: the water is seawater if its Sr/Ba ratio is greater than 0.8, mixed water if its Sr/Ba ratio is 0.5-0.8, and freshwater if its Sr/Ba ratio is less than 0.5^[12]. The Ni/Co element ratio is widely used to identify the redox environment: the environment is anaerobic if its Ni/Co ratio is greater than 7.0, oxygen-deficient if its Ni/Co ratio is 5.0-7.0, and oxygen rich if its Ni/Co ratio is less than 5.0. Elements Al and Ti are frequently used to indicate the input of terrigenous clastic materials. For the Shanxi Formation shale, the Sr/Ba ratio is 0.23-2.60, data projection points are located in the continental, transitional and marine regions (Fig. 6a), suggesting typical marine-continental transitional deposits. The Ni/Co ratio is 0.8-11.3 (Fig. 6b), indicating an anaerobic-oxygen-deficient-oxygen-rich sedimentary environment. C-L lies in the upper part of Shan₂³ sub-member, and belongs to tide-dominated deltaic facies. The Sr/Ba ratio of 0.23- 0.64 and the Ni/Co ratio of 0.80-4.65 point to a continental freshwater oxidized environment. Al₂O₃ content is 21.80% and TiO₂ content is 0.87% on average, indicating a sedimentary environment adjacent to the source. The input of a large amount of terrigenous clasts brought a large amount of argillaceous material, making the organic matter diluted, and resulting in high clay mineral content and low TOC. C-H lies in the mid-upper part of Shan₂³ sub-member, and belongs to the bay facies between deltaic distributary channels. The Sr/Ba ratio of 0.28-0.74 and Ni/Co ratio of 1.12-4.65 indicate a continental freshwater-mixed water oxidized environment. The average Al₂O₃ and TiO₂ contents are 25.90% and 0.99%, respectively, indicating a sedimentary environment adjacent to the source.

S-L, S-M, S-H, and M-M lie in the lower part of Shan₂³ sub-member and appear in the tide-dominated estuarine bay. S-L formed in a semi-closed lagoon environment. The Sr/Ba ratio of 0.38-0.49 and Ni/Co ratio of 3.20-4.69 indicate a continental freshwater oxidized environment. The average Al₂O₃ and TiO₂ contents of 17.4% and 0.66%, respectively, suggest a sedimentary environment relatively close to the source. S-M formed in a closed lagoon environment. The Sr/Ba ratio of 0.4-1.4 and Ni/Co ratio of 8.0-11.3 point to a transitional mixed water anaerobic environment. Closed lagoons are quiet with low-energy environments near oceans, where the relatively deep- water body is conducive to the preservation of organic matter. S-H and M-M formed in a bay environment. The Sr/Ba ratio of 0.8-2.6 and Ni/Co ratio of 7.7-9.4 suggest that the shale was deposited in an anaerobic marine media. The average Al₂O₃ and TiO₂ contents of S-M, S-H, and M-M are 10.5% and 0.4%, pointing to a closed lagoon-bay environment far away from the source, and less affected by terrigenous clasts. Thin sections reveal lenticular bedding (Fig. 4g, 4h), bearing numerous foraminifera, sponge spicules, and other siliceous bioclasts (Fig. 4l, 4m). The plankton died and sank to the seafloor, providing abundant organic matter and biogenic silica. It is noted that affected by both rivers and oceans during deposition, the organic sources of S-M, S-H, and M-M include both terrigenous vascular plants and marine plankton.

Differing sedimentary environments, mineral compositions, organic matter types and contents result in differing micropore characteristics.

The reservoir space of the Shanxi Formation shale include organic matter fractures, mineral interparticle pores, clay mineral interlayer pores, pyrite intercrystalline pores, and mineral dissolution pores. Different lithofacies types have different pore types (Fig. 7).

The C-L shale mainly contains clay interlayer pores, and a small amount of organic matter fractures and mineral interparticle pores (Fig. 7a-7c). On the SEM images, organic matter appears as bars with well-defined edges and corners, and the particles size ranges from a few microns to tens of microns. A small number of circular or oval primary pores of less than 20 nm in size have developed. Between the organic matter particles and the mineral particles, slender pores and fractures can be observed (Fig. 7b, 7c). The S-L shale contains mainly mineral interparticle pores, and a small number of primary pores can be seen in the bar-shaped organic matter (Fig. 7d, 7e).

The S-M, S-H, and M-M shales all contain organic matter pore-fractures and mineral interparticle pores, as well as clay interlayer pores, pyrite intercrystalline pores, and mineral dissolution pores (Fig. 7f-7n). The organic matter in these three lithofacies types exists in three different forms: clumpy, bar-shaped, and interstitial. The clumpy organic matter has slightly rounded peripheries, and the particles are up to ten microns. A large number of complex liquid hydrocarbon bubble pores were found, showing smaller pores nested in larger ones, with 50-200 nm in size (Fig. 7f-7m). Interparticle fractures are between the interstitial organic matter and the mineral particles, which are irregular and a few hundred nanometers in size (Fig. 7h, 7k). Numerous framboids pyrite aggregate are generally associated with organic matter, and contain many intercrystallite pores and interparticle pores of 50-100 nm (Fig. 7i), in clusters or dispersion. Furthermore, in the M-M shale, a large amount of microcrystalline calcite contains dissolution pores that are circular or irregular and approximately tens of nanometers (Fig. 7n). The C-H shale contains numerous clay interlayer fractures, and mold pores which are the result of the detachment of granular kaolinite (Fig. 7o, 7p).

Fig. 8a compares the low-temperature CO₂ isothermal adsorption curves of the six lithofacies types. As relative pressure increases, the adsorption rate increases too. The adsorption rate is the greatest when the relative pressure ranges from 0 to 0.01, denoting micropore-filling adsorption. C-L and S-L have the lowest adsorption rates, and their maximum adsorption rates are only up to 0.78 cm³/g and 0.76 cm³/g, respectively, micropore-specific surface areas are 2.18 m²/g and 8.48 m²/g, respectively, and mi-cropore volume is 0.003 cm³/g each. For S-M and M-M, the maximum adsorption rates are 1.75 cm³/g and 1.29 cm³/g, respectively, the micropore-specific surface areas are 21.38 m²/g and 18.18 m²/g, respectively, and the micropore volumes are 0.003 cm³/g and 0.006 cm³/g, respectively. S-H and C-H have the greatest adsorption rates, up to 2.99 cm³/g and 4.76 cm³/g, respectively, and their micropore-specific surface areas are 100.3 m²/g and 41.26 m²/g, respectively, and micropore volumes of 0.015 cm³/g and 0.033 cm³/g, respectively. This means that they have the most developed micropores.

Fig. 8b compares the low-temperature N₂ adsorption- desorption curves of the six lithofacies types. As relative pressure increases, the C-L adsorption curve rises rapidly, and its adsorption is still unsaturated even when the equilibrium pressure is close to the saturated steam pressure, denoting wedge pores with open ends or groove pores comprised of flaky particles. The mesopore-specific surface area and the pore volume are 5.44 m²/g and 0.017 cm³/g, respectively. The adsorption-desorption loop of S-L is broad, and the adsorption-desorption curve is largely sloped with rapidly declining inflections, suggesting pores with thin necks and wide bodies. The mesopore- specific surface area and the pore volume are 4.98 m²/g and 0.01 cm³/g, respectively. The S-M, M-M, and S-H adsorption-desorption curves are quite narrow in loop, rising gradually and then soaring steeply at high pressure, which reflects groove pores or slit pores with open sides. The mesopore-specific surface areas are 5.04, 3.12, and 2.36 m²/g, and the mesopore volumes are 0.012, 0.008, and 0.003 cm³/g, respectively. The C-H loop is the narrowest, with the mesopore-specific surface area and pore volume of just 0.47 m²/g and 0.003 cm³/g, respectively.

The morphology of the HPMI curves describes the structural characteristics of macropores. Fig. 8c compares the HPMI curves of the six lithofacies types. The HPMI curve of C-L first increases rapidly and then remains unchanged, suggesting that the macropores are quite developed. The mercury withdrawal efficiency of 40% suggests good pore connectivity. The HPMI curve of S-L rises stepwise with increasing pressure, and the pore throats are unevenly distributed. The mercury withdrawal efficiency is less than 5%, suggesting that a large amount of mercury is trapped in the pore system and not able to be released, indicating that the pore structures are complex and poorly connected. The S-M HPMI curve initially rises rapidly and then increases at a constant rate. The mercury withdrawal efficiency of 60% suggests good pore connectivity. The M-M HPMI curve first rises rapidly and then continues with a gradual increase. The mercury withdrawal curve first remains unchanged and then drops rapidly, and the mercury withdrawal efficiency is 30%. The S-H HPMI curve rises stepwise with increasing pressure, and the pore throats are unevenly distributed. The mercury withdrawal efficiency is as high as 60%, suggesting good pore connectivity. The C-H HPMI curve rises in line with low- and high-pressure intervals, while the mercury withdrawal curve remains almost unchanged. The mercury efficiency is lower than 5%, suggesting poor pore connectivity.

Overall, S-H has superior pore structural parameters, with a total pore volume of 0.023 3 cm³/g, a total specific surface area of 43.63 m²/g, and a porosity of 5.54%. S-M and M-M have relatively good pore structural parameters, with a total pore volume of 0.014-0.015 cm³/g, a total specific surface area of 21.3-26.4 m²/g, and porosity of 5.1%-5.4%. In the three lithofacies types, micropores account for 70% and mesopores + macropores account for 30%, promising good pore connectivity. In contrast, C-H has a total pore volume of 0.036 cm³/g and a total specific surface area of 100.77 m²/g. Micropores account for 94%, leaving little room for mesopores and macropores. C-L and S-L have inferior pore structural parameters, with a total pore volume of 0.013-0.020 cm³/g, a total specific surface area of 7.6-13.4 m²/g, and porosity of 0.58%-1.93%. Macropores account for 60%, while mesopores and micropores account for around 20% each (Fig. 8d).

The main criteria for identifying favorable lithofacies are organic matter type and abundance, the type and development of reservoir space, and type and content of minerals. Organic matter type and abundance determine the hydrocarbon generation potential of shales^[13]. Type-II kerogen, for example, has a better oil/gas generation potential than that of type III^[14]. Type and development of reservoir space affects shale reservoir performance^[15]. Brittle minerals determine the fracturability of the shale reservoir, the stronger the shale brittleness, the better the fracturing effect^[13].

Three favorable lithofacies, S-M, M-M, and S-H, developed in a shallow bay-closed lagoon environment (Fig. 5). The organic matter comes from lower marine organisms, plankton, and terrigenous plants, which constitute type-II₂ kerogens. As the environment is less affected by terrigenous input, richer organic matters accumulate in these lithofacies, with high TOC (5.5%-12.0%) and high core gas content (2.0-4.0 m³/t). In addition, they have diverse reservoir space types (Fig. 7), superior pore structural parameters, and good pore openness and connectivity. And of them, S-H has the best pore system development (Fig. 9a). These three lithofacies types develop clumpy, interstitial, and bar-shaped organic matters. The bar-shaped organic matter contains primary gas-generating pores^[15], which are simple in structure and dispersive in distribution, and less than 20 nm (Fig. 7c, 7e). Between the interstitial organic matter and the mineral particles are interparticle pores and fractures, a few hundred nanometers in size (Fig. 7l). The clumpy organic matter contains liquid hydrocarbon bubble pores^[16], which are complex in structure and distributed in clusters, generally more than 50 nm (Fig. 7h-7n). These three lithofacies types develop type-II kerogens, which have a pore development potential that is considerably greater than that of the type III kerogens^{[15, 16]}. During thermal evolution, type-II₂ kerogens first generate large amounts of liquid hydrocarbon, which is further cracked into gas, creating dense organic matter pores after the bubbles transformed into pores. FIB-SEM images and Avizo digital modeling results show that the type-II₂ organic matter contains large numbers of bubble pores that are highly spherical, with a primary pore size of 10-300 nm and porosity of approximately 6.5% (Fig. 10). Among the bubble pores, many isolated large pores developed, possibly constituting a multi-scale porous structure^[15]. The smaller pores contain adsorbed gas, while the larger pores contain free gas. These pores are connected to each other to form an effective three-dimensional pore network. JMicro Vision image analysis revealed that the areal porosity of S-M, M-M, and S-H is 1.00%-1.45%.

The C-L and S-L shale developed in a deltaic or semi-closed lagoon environment (Fig. 5). The organic matter comes from terrigenous plant debris, with the kerogens of type III. The input of large amounts of terrigenous debris has resulted in the TOC lower than 5.5%, and the core gas content of 0.7-0.9 m³/t. The C-H shale developed in an inter-distributary channel bay (Fig. 5), with TOC of 15%-25% and kerogen of type III. These three lithofacies types have simpler reservoir space, including mineral interparticle pores, clay interlayer pores and fractures, and a small amount of organic matter pores and fractures. The areal porosity is 0.35%-0.60%. These lithofacies types develop bar-shaped organic matter. The primary kerogen parent material is type III, which tends to generate gas. The output of liquid hydrocarbon is very small. The lack of bubble-to-pore transformation makes it difficult to generate large numbers of complex organic matter pores, and organic matter pores existed are small and poorly connected^[16,17] (Fig. 7c, 7e). However, the C-H shale has rich organic matters, and its micropores have high pore structural parameters (Table 1, Fig. 9). There should be numerous micropores in type III organic matter of the C-H shale, but they cannot be observed at low SEM resolution.

In the Shanxi Formation, S-M, M-M, and S-H have the richest brittle minerals up to 60%-80% (averaging 71%), and similar to the Longmaxi shale reservoirs in the southern Sichuan Basin in composition and content (averaging 70%), promising good fracturability. S-L has brittle minerals of 47%-70% (averaging 58%), and C-L has brittle minerals of 36%-52% (averaging 39%). These two types of lithofacies are less fracturable. C-H has brittle minerals of just 15%-35% (averaging 24%), giving it the lowest level of fracturability (Fig. 9b).

The S-M, M-M and S-H all have good and abundant organic matters, diverse reservoir space, superior pore structures and rich brittle minerals, making them favorable in the Shanxi Formation. The S-H shale has the best reservoir conditions, followed by S-M and M-M.

Favorable lithofacies of the Permian shale in the eastern margin of the Ordos Basin developed in the bay- lagoon sedimentary facies in the middle and lower Shan₂³ sub-member. They have high TOC, moderate R_o, good reservoir properties, high gas-bearing capacity, and high brittleness (Fig. 5). In Well Daji51, for example, the TOC of the favorable lithofacies interval is 6%-12%, the R_o is 2.4%-2.8%, the porosity is 3.0%-6.0%, the total pore volume is 0.015-0.025 cm³/g, the total specific surface area is 20-45 m²/g, the gas content is 1.5-2.5 m³/t, and the brittleness index is 60%-80%. In the favorable lithofacies interval, the post-fracturing gas production is stable at 0.6×10⁴ m³/d on average. During the production test, a total of 33.9×10⁴ m³ of shale gas was produced. Continuous production test on the tested interval lasted 1612 h, and the absolute open flow was up to 2.3×10⁴ m³/d. If production were allocated at 1/5-1/3 of the absolute open flow capacity, this well can effectively produce at 0.6×10⁴ m³/d. When compared with the average production of 0.8×10⁴ m³/d from the fractured vertical well in the United States, it is clear that this interval is worth developing^[7].

Sedimentary environment and diagenesis control the formation and distribution of the favorable shales in the Shanxi Formation. The sedimentary environment determines the mineral composition and structure of the shales^[18,19,20]. Favorable lithofacies of S-M, M-M and S-H were typically formed in a shallow bay-lagoon environment, and their facies are almost horizontal laminae. When they deposited, the water bodies were calm and the hydrodynamic conditions were weak. The high contents of dolomite and calcite indicate that the shale deposited in a fairly closed water body, as such, it was less affected by terrigenous debris. The Sr/Ba ratio of 0.8-2.6 and Ni/Co ratio of 7.7-11.3 indicate that the water body had a fairly high level of salinity and high reducibility, while the low Al₂O₃ and TiO₂ contents indicate limited terrigenous input. The lack of obvious correlation between the Al₂O₃ and TiO₂ contents and the SiO₂ content also demonstrates that terrigenous materials contributed to only part of the siliceous minerals. After comparing the Si and Al projection points of the Shanxi Formation shale with the organic-rich siliceous shales of the Longmaxi Formation in the Sichuan Basin and the Barnett Formation in North America (Fig. 11), it was discovered that the data points of the favorable shales in bay-lagoon are mostly located above the illite Si/Al line. This is similar to the Longmaxi and Barnett shales, suggesting a larger proportion of authigenic silica^[20]. The excess silica in the favorable lithofacies is approximate 25%-52% according to the formula in reference [21], denoting biogenic silica. In summary, a bay-lagoon environment has quiet water and sufficient sunlight for silica-rich plankton to thrive, therefore providing rich organic matter and biogenic silica. Closed and anoxic water and relatively low sedimentary rate is also favorable for burial and preservation of organic matter. All these factors combine to facilitate the formation of organic-rich siliceous-mixed shale.

Diagenesis controls the reservoir space of shales through organic and inorganic processes^[18]. Organic matters in favorable lithofacies consist of type-II₂ kerogens, which are rich in protein and lipid compounds, contain n-alkanes and n-fatty acids with medium relative molecular masses, and are rich in steroidal alkanes. These tend to produce large numbers of honeycomb-like organic matter bubble pores and forming a three-dimensional connected pore network^[22]. For non-favorable lithofacies, the organic matters consist of type III kerogens. These kerogens are rich in cellulose and lignin and contain n-alkanes, terpenoids, and hopane steroids, which have fairly high relative molecular masses and tend to generate gas when they are mature and produce limited numbers of organic matter pores^[22]. In the Shanxi Formation, the inorganic diagenetic processes primarily include compaction and cementation. In favorable lithofacies, the biogenic silica is formed in the early period of the early-middle diagenetic stages and constitutes a hard particle-supported framework, effectively preserving large amounts of primary pores during the subsequent compaction^[18,19]. Nanoscale biogenic silica is often closely associated with organic matter, offering important hosting space for early liquid hydrocarbon formation as well as space and protection for organic pores occurring at a high level of maturity and generation and storage of shale gas^[18]. Furthermore, favorable lithofacies contain calcite minerals, which are easily dissolved by organic acids to produce dissolution pores, contributing positively to reservoir improvement. In Fig. 11, the data points of Si and Al in non-favorable lithofacies in deltaic facies mostly fall below the Si/Al line. This suggests that a larger proportion of terrigenous clastic silica was transported into the basin by wind and rivers after the weathering of the parent rock. The particles are large, and have few primary pores, so they are weak against compaction. A few data points were found above the Si/Al line, suggesting that some of the samples also contain a small amount of quartz transformed from clay minerals. The quartz formed when the silica separated from montmorillonite which transforming into illite in the middle-late diagenetic stage^[23]. This type of cementation generally reduces the number of reservoir pores.

There are six types mianly of lithofacies in the Shan₂³ sub-member in the eastern margin of the Ordos Basin: low TOC clay shale (C-L), low TOC siliceous shale (S-L), medium TOC siliceous shale (S-M), medium TOC mixed shale (M-M), high TOC siliceous shale (S-H), and high TOC clay shale (C-H). C-L and C-H lie in the middle and upper parts of Shan₂³ sub-member and occur in tide-dominated deltaic facies; S-L, S-M, S-H and M-M lie in the lower part of Shan₂³ sub-member and occur in tide-dominated estuarine bay facies. S-M, M-M and S-H formed in a shallow bay-closed lagoon sedimentary environment, and they have good and abundant organic matters, diverse reservoir spaces, superior pore structures, and rich brittle minerals, which make them favorable lithofacies in the study area. S-H has the best reservoir conditions, followed by S-M and M-M. Sedimentary environment and diagenesis control the formation of the favorable lithofacies in the Shanxi Formation. A shallow bay-lagoon environment is conducive to the formation of type-II₂ kerogen, which tends to produce large amounts of honeycomb-like organic matter bubble pores. Rich biogenic silica favors the preservation of various pores and enhances reservoir fracturability.