Oil Retention in Shales: A Review of the Mechanism, Controls and Assessment

Shale oil is a vital alternative energy source for oil and gas and has recently received an extensive attention. Characterization of the shale oil content provides an important guiding significance for resource potential evaluation, sweet spot prediction, and development of shale oil. In this paper, the mechanism, evaluation and influencing factors of oil retention in shales are reviewed. Oil is retained in shales through adsorption and swelling of kerogen, adsorption onto minerals and storage in shale pores. Quite a few methods are developed for oil content evaluation, such as three-dimensional fluorescence quantitation, two-dimensional nuclear magnetic resonance (2D NMR), solvent extraction, pyrolysis, multiple extraction-multiple pyrolysis-multiple chromatography, logging calculation, statistical regression, pyrolysis simulation experiment, and mass balance calculation. However, the limitations of these methods represent a challenge in practical applications. On this basis, the influencing factors of the oil retention are summarized from the microscale to the macroscale. The oil retention capacity is comprehensively controlled by organic matter abundance, type and maturity, mineral composition and diagenesis, oil storage space, shale thickness, and preservation conditions. Finally, oil mobility evaluation methods are introduced, mainly including the multitemperature pyrolysis, 2D NMR, and adsorption-swelling experiment, and the influencing factors of movable shale oil are briefly discussed. The aim of this paper is to deepen the understanding of shale oil evaluation and provide a basis for further research.


INTRODUCTION
Compared to conventional petroleum systems, shale strata exhibit the characteristics of low porosity and ultralow permeability, and shale oil and gas have received insufficient attention for a long time due to the great difficulty of exploitation (Jarvie, 2012;Chen L. et al., 2019b;Soeder and Borglum, 2019). However, in recent years, with the progress of horizontal drilling, multistage hydraulic fracturing and other engineering techniques, shale oil and gas have become resources with an appreciable development potential (Wang et al., 2015a;Kumar et al., 2017;Zou et al., 2017;Zhao et al., 2018;Chen L. et al., 2019b;Ghanizadeh et al., 2020;Chen et al., 2021). The large-scale development of shale oil in the Permian Basin, Gulf Basin and Williston Basin has provided a new perspective for oil exploration and development . In 2019, the proven reserves in key shale oil exploration areas in the United States reached 232.40 × 10 8 bbl, shale oil production yielded 22.78 × 10 8 bbl (US Energy Information Administration, 2021), and shale oil exploitation is conducive to energy security and economic growth (Soeder and Borglum, 2020;Solarin, 2020). Russia, Argentina, Canada and other countries also possess a colossal shale oil resource potential, with recoverable shale oil resources of 746 × 10 8 bbl, 270 × 10 8 bbl, and 88 × 10 8 bbl, respectively (US Energy Information Administration, 2015). Inspired by the success of shale oil exploitation in the United States, these countries have launched shale oil exploration and development projects. The daily output of shale oil in Canada and Argentina is approximately 40 × 10 4 bbl/d and 5 × 10 4 bbl/d, respectively , the shale oil contained in the Mesozoic Bazhenov-Abalak Formations has been experimentally exploited in Russia, and the average horizontal well production reached 55.13 bbl/d in 2016 (Liang et al., 2019).
China contains abundant shale oil resources, the geological resources are estimated to be 27195 × 10 8 bbl , and the recoverable reserves of shale oil are 367.5 × 10 8 bbl (Zou et al., 2019b). With the deepening of petroleum exploration, shale oil has been found in the Ordos, Sichuan, Bohai Bay, Songliao, Junggar, Santanghu and other petroliferous basins, and notable exploration production has been documented from the Permian Lucaogou Formation in the Junggar Basin and Santanghu Basin, the Triassic Yanchang Formation in the Ordos Basin, the Cretaceous Qingshankou Formation in the Songliao Basin and the Paleogene Kongdian Formation in the Bohai Bay Basin (Wang et al., 2019a;Hu et al., 2020).
In contrast to conventional petroleum systems (oil migrates from source rocks to high-porosity reservoirs), shale oil is directly produced from shale strata and its associated fine-grained or organic-lean interlayers (Wang et al., 2019a). The more hydrocarbons are generated and retained, the higher the shale oil exploration potential (Wang E. Z. et al., 2021). Therefore, it is urgent to understand the oil and movable oil retained in shales. Although there have been many studies on shale oil, there is a lack of systematic investigation and summary on the mechanism of oil retention in shales and the influencing factors of oil-bearing properties.
Starting from the mechanism of shale oil retention, this paper compares various methods to quantify the total oil content in shales, summarizes its influencing factors from microcosmic to macroscopic perspectives, and then examines some available methods to estimate the movable oil in shales and discuss its influencing factors. It is anticipant that this paper could deepen the understanding of shale oil geology and provide a theoretical basis for shale oil resource evaluation and favorable exploration area prediction.
There may be a sizable capillary pressure causing oil to be drawn into and become stored in kerogen (Alfi et al., 2014). The dissolution of oil in kerogen causes kerogen volume expansion, which is similar to the swelling of organic polymers (Sandvik et al., 1992). The kerogen contained in an organic-rich shale exhibits a high cross-link density, and hydrogen bonds do not play a significant role in the intermolecular bonds of the network (Larsen and Shang, 1994;Larsen and Shang, 1997;Larsen et al., 2002;Ertas et al., 2006). The kerogen swelling behavior is controlled by the available free volume of kerogen and the size and shape of the adsorbed molecules (Stainforth, 2009). When hydrocarbon molecules are dissolved in kerogen, the interaction between the molecules and kerogen decreases, and the kerogen volume increases. Previous studies indicated that various solvents impose different swelling effects on kerogen (Larsen and Shang, 1997;Ritter, 2003). Therefore, as a multicomponent mixture, the swelling capacity of liquid petroleum depends on the contribution of each component (Pathak et al., 2017). Moreover, kerogen absorbs different components according to its preference, and oil components with similar values of Hildebrand parameter (δ) are more easily absorbed by kerogen (Ritter, 2003;Wei et al., 2012). The influence of kerogen swelling on the oil retention capacity of shales cannot be ignored. The swelling capacity of different types of kerogen varies. Tian et al. (2014) calculated that the swelling capacities of type I, II 1 and II 2 kerogens (Ro 0.5%) are 141.7, 119.2, and 94.9 mg hydrocarbons/g TOC, respectively. Adsorption of kerogen is an important retention mode of oil (pepper, 1991), in which active kerogen provides the most active adsorption sites (Han et al., 2015). Kerogen in shale can absorb oil on the basis of satisfying the swelling effect of oil , with the increase of maturity, the specific surface area of kerogen gradually change and are capable of adsorbing hydrocarbons on their surface . Tian et al. (2020) calculated that the adsorbed oil and the specific surface area of kerogen increased first and then decreased, the adsorbed oil content reaches its peak at Ro 1.0%.
The shale is dominated by nanosized pore system, therefore, the interaction of fluids and pores is prominent, in other words, the influence of the adsorption of mineral particles on shale oil retention is of importance Wu et al., 2021). Generally, the adsorption between oil and mineral surfaces is attributed to physical interactions (such as van der Waals forces, and hydrogen bonds) or chemical interactions (Yusupova et al., 2002;Dudášová et al., 2008;Cui and Cheng, 2017;. According to hydrocarbon-solution adsorption tests, it is believed that the asphaltene adsorption capacity of feldspar (7.0 mg/g) is higher than that of quartz (4.5 mg/g), which is related to differences in the crystal structure and chemical composition between quartz and feldspar (Ribeiro et al., 2009). The maximum adsorption capacity of calcite for extracted shale oil-asphaltene reaches only 2.16 mg/g (Mohammadi and Sedighi, 2013), and the adsorption capacity is relatively low, which may be related to the fact that calcite (a neutral mineral) generally exhibits no electric charge . Among inorganic mineral components, clay minerals provide the main adsorption surfaces for shale oil Ning et al., 2020).Due to the difference in cation substitution mechanism, charge and layer charge among clay minerals, the adsorption capacity of various clay minerals or even different surfaces within the same clay minerals for oil is different: 1) Al 3+ in the montmorillonite crystal layer is replaced by divalent cations to generate a negative charge, which can attract Ca 2+ or Na + , while Si 4+ ions in the illite crystal layer are located on the lattice surface, and the negative charge due to Al 3+ substitution generates a greater attraction force to K + so that illite can adsorb more negative organic macromolecules ; 2) kaolinite with an octahedral-tetrahedral structure, which is different from that of other clay minerals, contains oil-wet silicate surfaces and water-wet alumina surfaces, and the adsorption capacity of oil components per unit area is 2.47 and 1.44 mg/m 2 , respectively . Moreover, different types of clay minerals have various specific surface areas, which may also affect adsorbed oil content (Zhu et al., 2012). However, Li et al. (2016) obtained the oil adsorption capacity of clay minerals (18 mg/g), quartz (3 mg/g) and carbonate minerals (1.8 mg/g) based on oil adsorption experiments of shale minerals. Compared to the adsorption capacity of kerogen, the oil adsorption capacity of these minerals may not be dominant in organic-rich shales. Zhao et al. (2019) reported that only 5-10% of hydrocarbons are adsorbed onto the surface of mineral particles. Considering the water-bearing capacity of shale reservoirs, the oil adsorption capacity in the water-bearing state is mainly controlled by kerogen (the remaining oil per unit of TOC is approximately 179 mg/g), which may be related to the hydrophilicity of mineral particles in shales (Wang et al., 2019b).
Nano-and micron-sized pores and microfractures are widely developed in shale oil reservoirs, constituting the reservoir space of shale oil (Zou et al., 2013;Su et al., 2018;Jin et al., 2021). The occurrence states of shale oil in pores of different scales are distinct. Free oil occurs in microfractures (Cui and Cheng, 2017;Song et al., 2020) and large pores (for example, the threshold of free oil enrichment is 30 nm, based , such as intergranular, intergranular and dissolution pores. The oil contained in smaller nanoscale pores coexists in the free and adsorption states (Wang et al., 2015a). With decreasing pore size, the oil in pores increasingly exhibits the adsorption state (for example, the threshold is 5 nm, based on Wang et al. (2019b)).
For a sum up from above discussion, the swelling of kerogen, adsorption of kerogen and mineral particles, and pore and fracture storage are the critical reasons for shale oil retention ( Figure 2).

EVALUATION OF RETAINED OIL IN SHALES
Oil content in shales is one of the key indexes to evaluate shale oil enrichment. The characterization and evaluation of the shale oil content involve various methods, including the three-dimensional quantitative fluorescence method, two-dimensional nuclear magnetic resonance (2D NMR) method, extraction method, pyrolysis method, multiple extraction-multiple pyrolysis-multiple chromatography method, logging calculation method, statistical regression method, pyrolysis simulation experiment method and material balance method. These methods exhibit distinct characteristics ( Table 1). The three methods of the pyrolysis, thermal simulation experiment and material balance will be futher presented, which are commonly adopted to evaluate the shale oil conten based on geochemical data.

Pyrolysis Method
The free hydrocarbon content (S 1 ) based on Rock-Eval pyrolysis is considered to determine the oil content in shales quickly. However, light hydrocarbon losses inevitably occur in the sampling and experimental processes (Jarvie, 2012;Jarvie, 2014). The heavy components present in kerogen, micropores and disconnected pore systems in the adsorption state can only be released at higher temperatures or when the pyrolysis temperature of organic matter is reached (Jarvie, 2012;Li M. W. et al., 2020). Zink et al. (2016) proposed that the amount of extractable S 2 component is 2.2-3.6 times the measured S 1 amount, and if this part of highcarbon number alkanes and aromatic hydrocarbons is ignored, the shale retained oil content is much underestimated. Jarvie (2012) pointed out that the oil content can be obtained according to S 1 and S 2 before and after extraction as presented in Eq. 1. It formed tha basis to calculate the shale oil content. Han et al. (2015) found that the extracted S 1 (S 1x ) component mainly comprised the extraction solvent, which suggested that the amount of shale oil should be the sum of the S 1 before and the S 2 difference before and after extraction. However,  reported that S 1x might be isolated hydrocarbons in nanopores, which could be regarded as part of the adsorbed hydrocarbons. In order to describe the oil content of shales more accurately, some improved pyrolysis methods have been applied (Jiang Q. G. et al., 2016;Abrams et al., 2017;Romero-Sarmiento, 2019;Gentzis et al., 2021). These oil content evaluation methods do not consider the light hydrocarbon loss attributed to sample collection and preservation and experimental operation. Cooles et al. (1986) considered that light hydrocarbons (C 14-) accounted for 35% of the total oil, but most of these light hydrocarbons were lost. Jiang C. Q. et al. (2016) analyzed the light hydrocarbon loss of samples via Rock-Eval analysis and thermal desorption-gas chromatography-mass spectrometry (TD-GC-MS)/flame ionization detection (FID) and reported that the S 1 loss for organic-rich shales could reach 15% after 360 h of open storage. Chen J. Q. et al. (2018) found that the loss rate of hydrocarbon evaporation ranged from 11 to 89% when analyzing the shale of the Middle Permian Lucaogou Formation in the Jimusar Sag. Therefore, it is necessary to correct the light hydrocarbon loss for the accurate evaluation of the shale oil content (Beti et al., 2020). Based on light hydrocarbon loss correction, Abrams et al. (2017) added a S 1 loss term (Eq. 2) to the original oil content equation.
With increasing maturity, the light components in shales that are easily lost in the process of core preservation and experimental analysis, and the light hydrocarbon loss exhibits an increasing dynamic trend Li J. B. et al., 2020b). Based on the above reasons, according to the difference in pyrolytic hydrocarbons between sealed freezing treatment and conventional preservation treatment , newly proposed oil formation volume factor (FVF) (Chen Z. H. FIGURE 2 | Shale oil retention patterns. Shale oil is mainly in adsorption and free states (modified from Zou et al., 2013;Hu et al., 2021).
Frontiers in Earth Science | www.frontiersin.org August 2021 | Volume 9 | Article 720839 4 TABLE 1 | An overview of determination of retained oil content in shales.

Method
Operation step Parameter Characteristics Reference Three-dimensional quantitative fluorescence method Shale particles with a mass of M (g) are placed in a solvent with a volume of V (L). The immersion solution is collected and detected via quantitative three-dimensional fluorescence. The fluorescence oil concentration C (mg/L) can be converted into the fluorescence oil content C o (mg/g)  With the use of NMR logging and the factor analysis method, oil saturation index (OSI) and S1 are estimated by using the bin porosity (33 ≤ T 2cutoff < 80) corresponding to hydrocarbon production S 1NMR (120 × bp6 + 100) × TOC/100 Continuous oil content analysis data can be obtained, but the signals of the different fluids overlap, and the prediction results may contain errors Piedrahita    consider the recovery method of S 1loss at different maturities, the recovery coefficient of light hydrocarbon loss becomes larger as the maturity increases (Figure 3), which means the light hydrocarbon calibaration is necessary in the mature and highmature stage of shale.

Material Balance Method
The mass of organic matter in shales that can be converted into petroleum should be constant during thermal evolution if there is not material exchange with the external environment. According to this idea, Pang et al. (2005) proposed the hydrocarbon generation potential method of rock pyrolysis, which is a statistical method based on the principle of mass balance, to calculate the hydrocarbon generation, retention and expulsion of shales. For a specific shale sample, its hydrocarbon generation potential index (Q g 100 × (S 1 +S 2 )/TOC) should remain unchanged. The only reason for its change is hydrocarbon expulsion. Hence, the difference between hydrocarbon generation and expulsion is the amount of retained hydrocarbons in the shale ( Figure 4A). However, this method does not consider the data error caused by light hydrocarbon evaporation , and does not provide a clear limit on the recovery coefficient of the hydrocarbon generation potential. Chen and Jiang (2016) statistically fitted an empirical model of the hydrocarbon index (HI) and T max based on the measured data of shale samples (Eq. 3), converted this empirical relationship into the hydrocarbon conversion rate (Tr) (Eq. 4) and further clarified the hydrocarbon expulsion efficiency (Eq. 5). On this basis, Wang et al. (2020) thought that S 1 could be regarded as being wholly derived from S 2 via transformation, so the original hydrogen index (H 0 I ) was selected to define the maximum value of the hydrocarbon generation capacity of the different types of kerogen, and the hydrocarbon generation, retention and expulsion processes of shales were characterized, then the shale oil resource potential can be evaluated by the hydrocarbon retention potential (Q r ) (mg HC/g TOC) ( Figure 4B).
In the above previous methods, the determination of the hydrocarbon expulsion threshold depends on the inflection point of the hydrocarbon generation potential index, and the determination of the hydrocarbon expulsion threshold may not be accurate. Li C. R. et al. (2020) adopted a hydrocarbon generation statistical model to simulate the changes in the hydrogen index and Q g during the evolution of shales and proposed a quantifiable hydrocarbon expulsion efficiency (Er) (Eq. 6), and shale oil resources can be calculated by the difference between Tr and Er, which was applied to the Qingshankou Formation in the Songliao Basin and achieved a reasonable result ( Figure 4C).
where β and θ are variables related to the type of kerogen, c is the error correction factor, α is the weight fraction of carbon relative to the generated hydrocarbons, its average value is considered to be 83.3%, Q go (GPI o ) represents the original hydrocarbon generation potential index.

Thermal Simulation Method
Because it is challenging to collect shale sample from immature to overmature stages, the hydrocarbon generation, retention and expulsion evolution profiles of shales can be established through thermal simulation experiments . Thermal Frontiers in Earth Science | www.frontiersin.org August 2021 | Volume 9 | Article 720839 simulation systems can be divided into open, closed and semiclosed systems, and they produce different results (Ghanizadeh et al., 2020). Under geological conditions, petroleum generation, retention and expulsion follow a kinetic evolution process. In a semiclosed system, thermal simulation products are allowed to be discharged during generation, which may be more suitable for uses of geological evaluation Ma et al., 2018;Zhang D. W. et al., 2020). Thermal simulation results cannot be directly applied to the geological evaluation and prediction. According to simulation experimental data on hydrocarbon generation in low-maturity shales, kinetic parameters such as activation energy (Ea) and frequency factor (A) can be obtained, and the hydrocarbon generation, retention and expulsion characteristics of shales under geological conditions can be obtained via extrapolation of these parameters. A series of kinetic models of hydrocarbon generation, such as the overall reaction model, series reaction model and parallel reaction model, have been proposed. Wang et al. (2011) compared different kinetic models and considered that the discretely distributed parallel first-order kinetic model might be the most suitable model to describe the hydrocarbon generation process. At present, it is widely accepted that the thermal decomposition process of kerogen is similar to a series of independent parallel first-order chemical reactions (Wang et al., 2011;Chen et al., 2017). Although Stainforth (2009) suggested that the assumption of independent first-order reactions may be too simplistic, the prediction of a hydrocarbon generation kinetics model controlled by first-order chemical kinetics seems to be supported by actual maturity measurements (Peters et al., 2018). The vitrinite reflectance corresponding to different temperatures can be calculated with Easy%Ro (Sweeney and Burnham, 1990), PresRo ™ (Carr, 1999), Pa%Ro (Xiao et al., 2005), Basin%Ro (Nielsen et al., 2017) and Easy%RoDL (Peters et al., 2018) at various heating temperature and duration, combined with thermal simulation results, to determine the oil generation, retention and expulsion characteristics of shales at different maturation stages.

INFLUENCING FACTORS OF OIL RETENTION IN SHALES
Shale oil is the oil remaining in shales during oil generation and expulsion, which is influenced by various factors. To explore the influencing factors more clearly, we examines the influencing factors of oil retention from the microscale to the macroscale. Microscopically, the oil retention in TOC per unit weight is chosen as the subject, and the influence of the organic matter type and thermal maturity on the shale oil content is investigated. Semimicroscopically, the research object is the oil retention in unit rock weight and micron-to centimeter-scale deposition structure, and the effects of organic matter abundance, pore development and rock mineral composition on oil retention are compared. Macroscopically, the shale section is selected as the research object, and the influence of the shale thickness and preservation conditions on shale oil retention and resource evaluation is clarified.

Restriction of Organic Matter Properties on Oil Retention
Shales with oil potential mainly contains sapropelic organic matter, with a kerogen type of I or II, which exhibit a high effective organic carbon ratio and good oil generation potential (Chen and Jiang, 2015;Ma et al., 2018;Li M. W. et al., 2019b;Zou et al., 2019b;Zhao W. Z. et al., 2021). The kerogen type in shales exerts apparent control over their oil-bearing properties. Generally, type I kerogen retained more oil than type II kerogen (Bai et al., 2017;He et al., 2019). Sun J. et al. (2021), also vertified that the retained content of shales exhibits growing trend with the increase of their original HI within the "oil window" stage through thermal simulation experiments. However, some Frontiers in Earth Science | www.frontiersin.org August 2021 | Volume 9 | Article 720839 8 geological observations indicate that shales containing type II kerogen may contain more considerable shale oil resources compared to shales with type I kerogen Hu et al., 2018). This may occur because the better the type of organic matter is, the greater the hydrocarbon expulsion amount per unit of organic carbon Huang et al., 2017;Ma et al., 2018;Sun J. et al., 2021), resulting in a lower retainded oil content. Tian et al. (2014) analyzed shales in the Qingshankou Formation in the northern Songliao Basin and reported that for Ro > 0.9%, the oil content per unit mass of TOC of type II kerogen is higher than that of type I kerogen, which seems to be related to the faster reduction in volume of type I kerogen per unit mass of TOC ( Figure 5). In addition, a shale containing type II kerogen exhibits a high oxygen content, which facilitates the formation of organic acids, and the hydrocarbon expulsion efficiency is lower than that of a shale containing type I kerogen, thereby its shale oil content is higher .
The oil content in shales is relatively low at the immature stage. Nevertheless, if the thermal maturity is too high, this leads to the so-called secondary cracking of oil into gas in shales, and oil is not effectively retained . Exploration practices of shale oil revealed that the Ro value in the shale oil palys broadly ranges from 0.5 to 1.6% (Cardott, 2012;. Maturity affects the shale oil content based on the amount of oil generated and affects the hydrocarbon expulsion ability according to the composition of evolution products. With increasing thermal maturation, the density and viscosity of oil generally decrease, which encourages the flow and expulsion of oil (Wang et al., 2019b;Zhao W. Z. et al, 2021). The adsorptionswelling oil amount per unit of organic carbon is negatively correlated with increasing thermal maturity . When the oil storage space remains fixed, this inevitably leads to the expulsion of oil. Moreover, Panahi et al. (2019) proposed that the formation, opening of hydrocarbon expulsion microfractures and fluid expulsion rate are related to thermal maturity levels based on experiments. Therefore, a moderate degree of thermal evolution is conducive to shale oil enrichment.  pointed out that the amount of retained oil in shales with Ro values between 0.9 and 1.3% reaches a maximum ( Figure 5).

Control of TOC Content, Oil Storage Space and Mineral Composition on Oil Retention
Organic matter enrichment constitutes the basis of shale oil accumulation, and shales with a certain shale oil potential often exhibit a higher TOC content (Zou et al., 2013). When the type and maturity of organic matter are fixed, the TOC content determines the oil generation capacity of shales . Although a high TOC content suggests an increase in hydrocarbon expulsion efficiency , analysis data from shale oil plays and thermal simulation revealed that the TOC content is usually positively correlated with the shale oil content ( Figure 5) Han et al., 2015). Hou et al. (2020) noted that shales with a high TOC content still yield a notable advantage in oil retention due to their high oil generation efficiency. An increase in TOC content also increases the content of adsorbed oil Wang et al., 2019b). In addition, studies have demonstrated that a higher TOC content indicates that more organic acids are produced during the hydrocarbon generation process, which facilitates the development of dissolution pores and recrystallized intergranular pores Hu et al., 2019) and provides essential oil storage spaces . Therefore, an increase in organic matter abundance is conducive to shale oil enrichment. As shown in Figure 6A, the oil content per unit mass of rock becomes larger as the organic matter abundance increases.
Shales comprise terrigenous detritus, clay minerals, carbonate, pyrite and other inorganic minerals and organic matter (Chen L. et al., 2019a). Inorganic components (clay minerals, pyrite, etc.) affect the shale oil content through oil generation and retention capacity. Clay minerals can influence the hydrocarbon generation process and mechanism through a carbon ion mechanism and/or participate in the desorption and hydrogenation/decarboxylation of solid acids (Du et al., 2021;Song et al., 2021a). Transition group elements such as Fe, Co and V in shales impose a catalytic effect on organic matter hydrocarbon generation, which can generate more oil and gas in organic-rich shales at the early burial stage (Xi et al., 2020). Studies have shown that local cementation can reduce the discharge of shale oil via compaction inhibition, and carbonate minerals can react with organic acids to increase oil storage spaces Milliken et al., 2021). Although the wettability of shale surfaces exhibits mixed characteristics , clay minerals still provide considerable oil adsorption surfaces, and the increase in the clay mineral content is beneficial to the increase in oil content to a certain extent (Li Z., 2020). Furthermore, laminae are widely developed sedimentary structures in shales, representing the directional arrangement of organic matter or inorganic minerals. Shales containing laminae exhibit a high hydrocarbon generation ability, and shales containing fine laminae formed via biochemical deposition exhibit a more pronounced hydrocarbon generation ability Su et al., 2019). The interbedding of micron-to centimeter-scale laminae composed of different components reflects the microscale source-reservoir configuration of shales Liu B. et al., 2021). Shales with well-developed laminae often attain a higher oil expulsion efficiency Du et al., 2019) (Figure 6B). A considerable amount of oil migrates into feldspar-quartz laminae or carbonate laminae at the semi-to microscale after meeting the oil retention requirements of organic-rich laminae, resulting in relatively high oil-bearing properties (Xi et al., 2020;Han et al., 2021).
Oil is adsorbed onto kerogen and mineral particles and occurs in pores and fractures in shales. The development of shale pores exerts an important control on the oil content   (Figure 6C). Chen G. H. et al. (2018) reported that 80% of shale oil is distributed in macropores. In fact, with increasing pore size, pore volume and porosity, the oil content in shales usually increases (Wang et al., 2019b;Song et al., 2020;Wang B. Y. et al., 2021). Because the density of the generated oil and gas is lighter than that of kerogen, the volume expansion and the overpressure phenomenon occurs in shales, resulting in the generation of oil expulsion fractures (Sun L. D. et al., 2021). Although shale microfractures form effective oil migration channels and oil storage spaces, when the scale of microfractures reaches a certain extent, this may reduce the shale oil content. Liu B. et al. (2021) pointed out that a higher fracture density resulted in a vertical migration of shale oil, which does not promote shale oil enrichment.  indicated that natural fracture development limited to the shale interior is relatively favorable to the seepage and preservation of retained oil in shales.

Influences of Shale Thickness and Preservation Conditions on Oil Retention
The thickness of organic-rich shales macroscopically controls the shale oil content. A thick shale, especially at its center part, does not facilitate oil expulsion (Leythaeuser et al., 1984). It seems that there is an effective petroleum expulsion thickness in geological conditions, and the shale beyond the thickness would have a weak oil expulsion (Jiang et al., 1986;Wang et al., 2005;Hou et al., 2017). Based on the pyrolysis experiment and geological model, Hou et al. (2017) found that the increase of the shale thickness during the oil generation stage would significantly reduce the oil expulsion efficiency. The thinner a single shale seam in a shale-FIGURE 6 | (A) Relationship between oil content and TOC of organic-rich shale (data from Wang M. et al., 2015;Li X. N. et al., 2017;Hu et al., 2018;Xu et al., 2019;Li C. R. et al., 2020). (B) relationship between lamina development and oil expulsion rate (modified from Wang Y. et al., 2016) (C) relationship between oil content and pore volume (modified from Zhao X. Z. et al., 2020).
Frontiers in Earth Science | www.frontiersin.org August 2021 | Volume 9 | Article 720839 sandstone interbedding stratum, the more frequent the shalesandstone configuration and the lower the retention of oil in the shale (Fan and Shi, 2019), while the oil-bearing properties of the interlayers (i.e., the sandstone seams) within the shale seams are improved (Raji et al., 2015). Exploration practices have verified that fault systems impart complementary control effects on both conventional and shale reservoirs, the retained oil in the shale may migrate to the shallow layers along the faults remaining active, which is unfavorable for shale oil enrichment Liu B. et al., 2021). However, thick shales are not easily broken through . Microfractures extending only in the interior of a shale increase the shale oil storage space and fluidity, which is more benefit to shale oil enrichment . Therefore, the shale thickness plays a positive role in the oil-bearing capacity (Fan and Shi, 2019;Zhao W. Z. et al., 2021). Preservation conditions are also important for shale oil enrichment. Lin et al. (2020) calculated that the decisive tectonic action in the Miquan area of the Junggar Basin caused a loss of 78.9% of the original shale oil resources. A relatively stable tectonic environment and high-quality top and bottom seal layers of a shale seam play positive roles in the preservation of shale oil Li M. W. et al., 2019b;Liu B. et al., 2021).

EVALUATION AND INFLUCEING FACTORS OF MOVABLE OIL IN SHALES
The movable oil content in shales is the key to a high yield of shale oil (Jarvie, 2012;Li M. W. et al., 2019a;Hu et al., 2021). It is highly important to determine the amount of movable oil and its influencing factors for shale oil prediction. This section mainly introduces three feasible and realistic methods to evaluate the shale movable oil content, including the pyrolysis, 2D NMR and adsorption-swelling calculation methods. In addition, the influencing factors of movable oil in shales are also discussed.

Pyrolysis Method
According to available empirical data of shale oil production, Jarvie (2012) proposed the oil crossover effect, i.e., OSI > 100 mg hydrocarbons/g TOC as the indicator of movable oil in shales ( Figure 7A). Quite a few authors have applied this standard to evaluate movable shale oil because it eliminates the influence of organic matter adsorption Cao et al., 2017;Hu et al., 2018;.  indicated that the movable oil content is equal to the difference between the amount of retained oil and the oil crossover effect value ( Figure 7B). However, the OSI method may not be fully applicable in lacustrine shales Huang et al., 2020), and the OSI value and oil crossover effect dynamically change with the maturity. With increasing maturity, the OSI values gradually increase until a maximum retention capacity is reached, and then a subsequent decrease is show ( Figure 7C) (Han et al., 2017).
The light components contained in macropores and connected pore systems in the free state can be released at lower temperatures. In contrast, the heavy components in kerogen, and micropores and disconnected pore systems in the adsorption state can only be released at higher temperatures or when the temperature of organic matter cracking is reached (Jiang Q. G. et al., 2016). Therefore, the amount of movable shale oil can be evaluated by the thermally released hydrocarbon peak obtained under the constant-heating rate mode. Maende (2016) noted in the HAWK Petroleum Assessment Method that the movable oil content is the sum of Oil-1, Oil-2, Oil-3 and Oil-4 ( Figure 8A). Jiang Q. G. et al. (2016) and Li J. B. et al. (2020a) considered that the thermally released hydrocarbon peak (S 1-1 ) at 200°C could represent the actual movable oil content, and the maximum movable oil content is the sum of S 1-1 and S 1-2 (at 350°C) ( Figure 8B). Romero-Sarmiento (2019) proposed that sorbed liquid hydrocarbons is adsorbed on the residual solid OM, and thus the movable oil content of shales can be determined according to the area difference of thermal peaks (Sh0 (at 100-200°C) and Sh1 (at 200-350°C)) of whole rock sample and the corresponding isolated OM (Figures 8C,D). Chen Z. H. et al. (2019) and Li M. W. et al. (2020) applied the thermally released hydrocarbon peak (S 1a ) at or below 300°C to represent the residual free hydrocarbons ( Figure 8E) and believed that the shale oil discounted by the critical saturation is not limited by adsorption forces and nanopore bonds but exhibits fluidity (Eq. 7). Gentzis et al. (2021) thought that L1, L2 and a part of L3 represented movable oil, whereas the remaining L3 and L4 containing heavy n-alkanes, resins, asphaltenes and NSO compounds represented immovable oil ( Figure 8F). Multistep pyrolysis method has low cost and easy operation, but it should be pointed out that the recovery of light hydrocarbon loss is necessary.
where S 1c is total free hydrocarbons after light hydrocarbon correction, TOC represents the amount of adsorption discount (mg HC/g TOC), which is equivalent to the percentage content of TOC in the sample.

NMR Method
A shale is characterized by complex porous media and multiple fluid occurrences (oil, gas, water and kerogen). The standard T 2 relaxation time distribution obtained by 1D NMR provides limited information, and signal overlap between the different fluids is severe (Birdwell and Washburn, 2015). Although T 2 -D can be used to distinguish the fluid phase (oil, gas and water) in shales, the solid-like organic matter signal and kerogen signal in micro/nanopores cannot be effectively distinguished Bai et al., 2019;Song and Kausik, 2019). In contrast, 2D NMR, especially high-frequency 2D NMR, can improve the detection ability of trace signals and can more efficiently distinguish light oil, heavy components, kerogen, bound water and other fluids in shales. Specifically, kerogen (with high T 1 ) and structural water have shorter T 2 and wider T 1 , oil generally has a higher T 1 /T 2 ratio than water (with low T 1 ), and the T 2 value of adsorbed oil is lower than that of movable oil Khatibi et al., 2019;Song and Kausik, 2019;Zhang P. F. et al., 2020) ( Figure 9A). Based on the principle that 2D NMR can effectively distinguish hydrogen nuclei of fluids in different occurrence states, researchers have combined 2D NMR with geochemical experiments, such as rock pyrolysis, solvent extraction, and quantitative grain fluorescence on extract, to comprehensively analyze the content of movable shale oil Liu B. et al., 2019;Li J. B. et al., 2020a;Gentzis et al., 2021). For example, Li J. B. et al. (2020a) extracted organic hydrogen T 2 spectra of kerogen, adsorbed oil and free oil at T 1 intervals corresponding to the original samples, samples pyrolyzed at 350°C and extracted samples. Their accumulative amplitude determined the T 2 cutoff values of free oil and adsorbed oil, and a nondestructive method for the direct evaluation of the free oil content was then provided ( Figures  9B,C). However, it should be noted that the threshold value of the T 1 -T 2 spectrum is related to the echo interval and frequency of the instrument itself (Nicot et al., 2016;Song and Kausik, 2019). The temperature and state of the samples and the magnetic minerals in shales would change the distribution of signal response interval . Furthermore, it is difficult to define the boundary of the T 1 /T 2 ratio of movable and nonmovable fluids, which may lead to uncertainty in movable oil evaluation. Overall, for different testing method, they both have advantages and limitations, for accurate desripting movable oil content, various methods should be using synthetically.

Adsorption-swelling Calculation Method
The shale oil adsorbed onto minerals and kerogen is usually immovable, and the movable oil content can be regarded as the difference between the total oil content and the adsorption-swelling oil content. Because the adsorption capacity of kerogen is much higher than that of minerals for shale oil, previous studies have mainly focused on the calculation of the adsorption and swelling capacities of kerogen (Wei et al., 2012;Tian et al., 2014), but this has caused an overestimation of the movable oil content. Therefore, Li et al. (2016) proposed the adsorption potential S p model (Eq. 8) to calculate the oil adsorption amount in shale, and the difference between S p and S 1 is suitable within a broader range of TOC and maturity levels . The adsorption-swelling oil quantity (Q a-s ) can also be calculated via molecular dynamics simulations (Eq. 9) Tian et al., 2020). The swelling capacity of kerogen (Q S ) is calculated according to the shale oil density curve for kerogen. According to the density curve and specific surface area of shale oil in kerogen and mineral pores, the adsorbed oil quantity (Q a ) can be deduced. However, the adsorption-swelling calculation method is primarily a simplified model of shale under geological conditions. There may be differences between the premise and assumed parameters and the geological conditions, which results in insufficiently accurate calculations of adsorption-swelling oil content. Optimization of this model is a future research direction.
Lm2 S mo ρ mo dL (9) where p O and x O are the proportion and adsorption capacity of OM respectively; p i and x i are the proportion of the ith mineral and its adsorbed oil amount respectively; p o + n i 1 p i 1; φ and φ o is porosity and shale initial porosity respectively; L o1 and L o2 are the starting and ending positions of the intersection of kerogen density curve and shale oil density curve respectively; L k1 and L k2 are the starting point and end of kerogen adsorption layer respectively; L m1 and L m2 are the starting point and end of mineral component adsorption layer respectively; S ko is the cross-sectional area of swelling and adsorption model of shale oil in kerogen, ρ ko is the density curve of shale oil, S mo is the adsorption model area of shale oil in mineral pores, and ρ mo is the density curve of shale oil in mineral pores.

Influencing Factors of Movable Oil
Great effort has been made to explore the influencing factors of the shale oil mobility through various means of production and experimental analysis methods. It is recognized that occurrences of retained oil in shales have a great influence on its fluidity. Adsorption-swelling oil occurs in the form of high-density solidlike or embedded kerogen on the surface of organic and inorganic pores (Wang et al., 2015b), and it is immovable without additional forces . The shale organic matter abundance exerts two impacts on the shale oil mobility. Even considering shale hydrocarbon expulsion, a high TOC is still beneficial to an increase in the shale oil and movable oil contents Hu et al., 2021;Zhao W. Z. et al., 2021). However, in shales with the same oil content, an increase in the TOC content suggests a decrease in the movable oil content (Wang et al., 2019b). The relationship between the organic matter type and movable oil indicates that a higher oil generating capacity per Frontiers in Earth Science | www.frontiersin.org August 2021 | Volume 9 | Article 720839 unit of organic carbon means a higher movable oil content (He et al., 2019). It has been reported that the proportion of movable oil will increases with increasing burial depth. This phenomenon is related to the decrease in oil viscosity and density due to increasing temperature and light components (Wang et al., 2019b;Scheeder et al., 2020;Zhao W. Z. et al., 2021), which is conducive to a reduction in the oil film thickness and a relative increase in free oil in pores with the same pore size . The increase in thermal maturity also leads to a decrease in the adsorption and swelling capacities of kerogen Wang et al., 2015b;Wang et al., 2019b).  reported that the decrease in adsorption capacity could be attributed to the reduction in TOC content with increasing maturity. Wang et al. (2015b), based on molecular dynamics simulation analysis, proposed that with increasing maturity, the number of carbonyl in kerogen gradually decreases, which leads to the weakening of the adsorption capacity of kerogen for alkanes.  conducted thermal simulation and kerogen swelling experiments on Paleogene shales obtained from the Huanghua depression of the Bohai Bay Basin and observed that an middle thermal maturity is the superior evolution stage for movable oil enrichment in lacustrine shale, which was attributed to the optimum matching of hydrocarbon generation and organic matter adsorption-swelling. Mineral particles and composition play specific roles in controlling movable shale oil. Overall, although intergranular pores can be formed among clay minerals, tiny intergranular pores and the high surface adsorption capacity of clay minerals hinder the storage of movable liquid hydrocarbons (Su et al., 2018;Ning et al., 2020). The adsorption affinity of quartz and carbonate minerals for oil is low (Raji et al., 2015;Li et al., 2016), and the increase in their contents promotes the development of microcracks (Ougier-Simonin et al., 2016) and the preservation of macropores , which facilitate the occurrence of mobile liquid hydrocarbons. However, Hu et al. (2021) analyzed the lacustrine shale of the Paleogene Shahejie Formation in the Dongpu Sag of the Bohai Bay Basin. Their results revealed that the movable oil content first increases and then decreases with the content of clay minerals (25% as the inflection point) and quartz minerals (15% as the inflection point), which depends on the coupling effect of the mineral content on the porosity, adsorption capacity and TOC content. In addition, due to the petroleum fractionation effect, feldsparquartz or carbonate laminae interbedded with organic-rich laminae at the microscale and interbedded shale, sandstone, carbonate, and mixed rock layers in shale strata at the macroscale contain more light oil components that easily flow (Jarvie, 2014;Fan and Shi, 2019;Pan et al., 2019;Liu B. et al., 2021).
Shale reservoir conditions are fundamental to the flow ability of shale oil. Bao (2018) speculated that shales within the oil generation window and a porosity higher than 6.5% promoted the storage of movable oil through the relationship between the OSI and porosity for the Paleogene shale in the Dongying Sag. The shale pore structure is a crucial factor controlling the shale oil mobility (Davudov et al., 2020;Jiang et al., 2020;Liu et al., 2020;Ning et al., 2020). Zou et al., 2015 found that oil cannot flow through pores below the lower limit of the flow threshold (20 nm) through simulations of a nanoporous template with a controllable pore diameter. Wang et al. (2015a), through adsorption simulations of alkanes onto a graphene surface (oil-wet), proposed that alkanes only flowed after exceeding the critical pore size. Conversely, they all occurred in the adsorption state. The pore size of shale reservoirs should exceed the flow threshold, which is a necessary condition for movable shale oil. The proportion of movable oil gradually increases with increasing pore size  because the pore connectivity of macropores is higher and provides a considerable pore volume for movable oil (Li T. W. et al., 2017;Jiang et al., 2020;Ning et al., 2020), and the diffusion coefficient of oil increases with increasing pore diameter Zhang et al., 2019). When the pore throat radius is large, the diffusion motion of FIGURE 9 | Determination of shale movable oil and adsorbed oil according to T 1 -T 2 spectrum (modified from Li J. B. et al., 2020a). AR is the original sample, P350 is the sample after pyrolysis to 350°C, EX is the sample after solvent extraction, T 2CF is the T 2 cut-off value of movable oil, T 2CA is the T 2 cut-off value of adsorbed oil.
hydrocarbon molecules of different sizes in throats is not restricted by the diffusion energy barrier . Cui and Cheng (2017) suggested that movable shale oil was also related to the porosity and pore size of organic pores, and the movable oil content decreased with increasing organic porosity ratio. The porosity and permeability of shales are low, and fractures can effectively improve the seepage capacity of shale reservoirs (Soeder and Borglum, 2019). Microfractures in shales are conducive to increasing the shale movable oil proportion and effective production . Moreover, flow experiments have demonstrated that the shale oil fluidity is related to the fracture strength and direction, and bedding microfractures play an essential role in shale oil migration .

CONCLUSION
(1) The oil generated from organic-rich shales can be partly retained in themselves via through adsorption and swelling of kerogen, adsorption onto minerals and storage in pores. These mechanisms have been applied to explain to oil retention in shales under geological conditions, while the relative thermal dynamics is looked forward to further development.
(2) Quite a few methods have been developed to characterize and evaluate the oil in shales. The pyrolysis, thermal simulation experiment and material balance methods are most common use to quantify the oil content in shales.
(3) The shale oil retention capacity is bascially controlled by the organic matter abundance and properties (type and maturity), mineral composition, oil storage space, shale thickness and preservation conditions, but inconsistent recoginitions exist, and are mainly from different scale research. It is necessary to strengthen a comprehensive study from the microscale to the macroscale. (4) The study on the movable oil in shales still remains at the exploratory stage.
Although a few method are believed to be effective to estimate the movable oil content in shales, their limitations indicate needs for the supplement and verification of geological data. Especially, the influencing factors of shale movable oil are complex and variable. More accurate evaluation parameters and correction methods on this topic will become a research focus.