The samples were crushed to 180–200 mesh prior to geochemical analysis. The total organic carbon (TOC) content was tested by a LECO CS-344 carbon and sulfur analyzer. The type, maturity, and hydrocarbon generation potential of OM were determined. The temperature at which the most S2 is generated is regarded as the maximum pyrolysis yield temperature (Tmax; ∘C), and the hydrogen index (HI; mg HC/g TOC) was obtained from the ratio of S2 to TOC (Peters, 1986).

The samples were tested by a D/Max 2500 diffractometer, and the tests follow two separate processes of the CPSC procedure (CPSC, 2010). The actual mineral content of the samples was obtained by quantitative X-ray diffraction (XRD) analysis of samples smaller than 200 mesh (<0.075 mm).

The bulk shale samples were uniformly crushed into 60 mesh with a laboratory knife mill and then evenly divided into five parts. The sample was extracted using solvents of different aromaticity, polarity, and permeability, namely acetone, tetrahydrofuran, carbon disulfide, and benzene; the remaining original sample was used as a control. About 1000 g solvent and 50 g shale sample were sealed in a Soxhlet extractor to prevent solvent loss. The extraction time lasted 150 h until the solvent became colorless. Samples were named according to their corresponding abbreviated well names (JH2, J1, LH10, and H317; Table 2), and the original shale samples were marked by the shorthand subscript, following the sample name (JH2orig, J1orig, LH10orig, and H317orig). The residue (e.g., JH2) from tetrahydrofuran (THF), acetone, CS2, and benzene extraction was clearly characterized by JH2T, JH2A, JH2C, and JH2B, respectively.

Both powdered original shale samples and their corresponding extracted shale residues (∼2 mg) were blended with potassium bromide (KBr; ∼200 mg) and pressed into pellets. Through the use of a Nicolet 6700 FT-IR spectrometer collecting 200 scans per sample at a resolution of 4 cm-1, the vibration frequencies of specific atomic bonds were analyzed. Thus, the chemical functional groups could be assessed using a standard semi-quantitative technique. For the KBr–FTIR characterization, each spectrum was recorded in the spectral range of 400–3250 cm-1. The bands of absorption were identified by comparing the spectrums with published studies (Drobniak and Mastalerz, 2006; Furmann et al., 2013).

According to a manual mode, the band area was determined using Origin software. These integrated band areas were used to calculate ratios for the (1) aromaticity, including AR1 = [(δArC = C/νAlC-H) + δArC = C], AR2 = (νArC-H/νAlC-H), and AR3 = (γArC-H/νAlC-H); (2) length of aliphatic chains, namely LAC = (CH2/CH3) in νAlC-H; and (3) degree of substitution (DOS) of aromatic sites by alkyl groups, namely DOS = (γArC-H(1 adjacent H)/γArC-H (4 adjacent H); Wei et al., 2014). The detailed definitions of the parameters can be found in Tables 1 and 4.

The representative spectrums of the original shale samples and corresponding residues are presented in Figs. 5 and 6, respectively. In terms of the length of aliphatic chains (LAC), longer n-alkyl chains are determined by CH2 bands, whereas shorter n-alkyl and branched aliphatic chains are indicated by CH3 bands (Table 4). The DOS of aromatic rings represents the fraction of aromatic sites where hydrogen atoms are substituted by alkyl groups (Table 4).

The geochemical features of shale samples are listed in Table 2. Four selected samples span from marginally mature (JH2; 439 ∘C) to late mature (H317; 460 ∘C), based on the Chinese continental hydrocarbon source rock geochemical evaluation method (SY/T 5735-1995, 1995). The kerogen type of selected samples is type I or II1 (Fig. 3; Tong et al., 2011). TOC content is generally high, ranging from 3.79 wt % to 22.19 wt % (Table 2). The hydrocarbon generation potential (S1+S2) ranges from 8.11 to 127.42 mg g-1. Table 2 shows that the hydrogen index (HI) of the samples is in the range of 155–528 mg g-1.

The samples contain a large number of clay minerals (42.8 wt %–54.2 wt %, with an average of 48.15 %). An illite–smectite mixed layer or illite are the most dominant clay types (Table 3). In general, the content of the felsic component is between 19.5 wt % and 52.2 wt %, with the average of 40.63 %. In addition, the pyrite content of the studied samples varies greatly, ranging from 0.9 wt % to 25.5 wt % (Table 3).

The extraction data of the studied samples with different solvents are shown in Fig. 4. Among all the solvents, THF has the highest extraction efficiency (2.44 wt %–4.43 wt %; Fig. 4a). According to the TOC-normalized yield of extracts, the mature samples J1 (Tmax+449∘) had significantly higher extractable samples (13.48 wt %–56.91 wt %) than other samples (≤5.28 wt %; Fig. 4b).

The representative spectrums of the original shale samples and corresponding residues are presented in Figs. 5 and 6, respectively. The FTIR results after extraction are summarized in Table 4. The OM in shale is a three-dimensional macromolecule composed of aromatic clusters and aliphatic chains (Tong et al., 2011; Guan et al., 2015; Wang et al., 2016, 2017; Zhao et al., 2018). The aromaticity (i.e., AR2 and AR3) of the samples follows the order H317 (3.327) > LH10 (3.225) > J1 (3.038) > JH2 (2.828; Table 4), which is consistent with maturity. In addition, shale residues from THF extraction possess higher aromaticity than those from other solvents. Unlike aromaticity, the length of aliphatic chains (LAC) does not increase monotonically with the increase in the maturity (Table 4). However, there is no regularity in the DOS (i.e., the fraction of aromatic sites where hydrogen atoms are substituted by alkyl groups) of different maturity shale samples, and the DOS of shale before and after extraction with different solvents demonstrated an irregular change (Table 4).

Previous studies have found that the extraction yields of shale and the composition of the soluble mixture are related to the properties of the solvent, extraction temperature, the geochemical parameters of the shale, and so on (Hu et al., 1999; Olukcu et al., 1999; Shaohui et al., 2000; Abourriche et al., 2013; Cao et al., 2019). As the polarity of the solvent increases, the extraction yield increases accordingly (Masaki et al., 2004). The solvent dissolution capacity has obvious temperature dependence, and the dissolution capacity increases with the increase in the temperature (Hu et al., 1999). The increase in extraction pressure and temperature will promote the activation energy of the sample (Allawzi et al., 2011).

The Hansen solubility parameters of the four solvents are shown in Table 5; the extraction ability of CS2 is similar to that of benzene (Table 5). Because of the aromatic structure, high boiling point, and strong polarity, THF could have a stronger extraction ability than other solvents (Table 5).

Allawzi et al. (2011) found that the OM obtained by solvent extraction is mainly composed of low molecular weight compounds (e.g., saturated hydrocarbons, olefin hydrogen, and aromatic compounds). Our results show that the extraction yield of shale with THF is significantly higher than that of other solvents (Fig. 4a), which could be attributed to the fact that a mass of the heaviest macromolecular compound of asphaltene was screened out (Cao et al., 2020). The significant elimination of OM (i.e., asphaltene) by THF may be related to the properties of the THF, including the aromatic structure (Johnson et al., 1967), high boiling point (Cao et al., 2019), excellent Hansen solubility parameters, and strong polarity (Toolan et al., 2016; Table 5), which could assist to destroy the interact between shale matrix and the asphaltenes (i.e., a complex molecular architecture with a central polycyclic aromatic hydrocarbon (PAH), with peripheral alkane substituents; Schuler et al., 2017) of PAH molecular polymerization (Schuler et al., 2015).

The extraction yields of the studied samples are of the order of LH10, JH2, H317, and J1, regardless of solvents (Fig. 4a), which may be due to the soluble hydrocarbon content (S1) of the studied sample (Table 2). According to the TOC-normalized yield of extracts, the mature samples J1 (Tmax=449∘) had significantly higher extractable samples than other samples (Fig. 4b), which may be related to the fact that the J1 sample (Tmax=449∘) was at the peak of hydrocarbon generation. Thus, a large number of kerogens were cracked into oil and bitumen (Ertas et al., 2006; Wei et al., 2014). However, a little kerogen was converted to a small amount of oil in the early maturity stage (Cao et al., 2020), so the TOC-normalized extraction yield of the JH2 sample is low. However, the TOC-normalized extraction yield of late mature shale H317 is similar to that of early mature shale JH2, which could be attributed to the fact that some of the oil is cracked to natural gas or transported to other locations with maturation (Cao et al., 2020).

The application of FTIR spectroscopy provides an evaluation of the aromaticity and length of aliphatic chains for original shale samples and residues from extractions. Table 4 showed that the aromaticity (AR2 and AR3) of the original shale OM with different maturity is obviously different and ranges from 2.828 to 3.327. Figure 7 shows there is a positive correlation between aromaticity and maturity in shale (Feng et al., 2013). Thus, the abovementioned facts indicate that the thermal maturation can increase the aromaticity of shale OM.

Table 4 showed that the length of aliphatic chains (LAC) from late mature shale (i.e., LH10 and H317) is higher than those of mature shale J1 and marginally mature shale JH2. In addition, the LAC of the early mature shale J1 is also higher than the marginally mature shale JH2 (Table 4). However, the LAC of the higher-maturity H317 shale (Tmax=456∘) is lower than that of mature LH10 shale (Tmax=454∘). Those values indicate that the aliphatic chains lengthened with the increase in the maturity before the late maturity stage (Feng et al., 2013) and then shortened as the maturity further increases. That may be attributed to the fact that kerogen cracking to form a large amount of bitumen at the low-maturity stage leads to longer aliphatic chains with increasing maturity, while at the high-maturity stage, bitumen is largely consumed, and the OM represented by bitumen is further cracked, thus shortening the aliphatic chains with increasing maturity (Feng et al., 2013). However, previous studies have also observed that aliphatic chains lengthened become monotonically shorter and/or more branched with maturity, which indicates that longer aliphatic chains can be preferentially cleaved. The cleavage of the C–C bonds between carbons α and β to a ring could explain this result (Lin and Ritz, 1993). Moreover, the large differences in the sources of OM in different regions may be the cause of the above contradictions.

The DOS of aromatic rings represents the fraction of aromatic sites, where hydrogen atoms are substituted by alkyl groups, and there is no regularity in the DOS (i.e., the fraction of aromatic sites where hydrogen atoms are substituted by alkyl groups) of different maturity shale samples (Table 4). Therefore, the fraction of aromatic sites where hydrogen atoms are substituted by alkyl groups in shale is independent of maturity.

In addition, the FTIR-derived LAC values smaller than 1 show the superiority of CH3 bands (i.e., asymmetric stretching vibration) in shale samples corresponding to short and branched aliphatic chains (Furmann et al., 2013). Our results demonstrate that aliphatic chains are shorter with higher branching in the original shale samples, which could not be changed during the extractions (Table 4). In addition, the FTIR of shale samples after different solvent extractions is irregular, which may be influenced by the inorganic minerals. Therefore, it seems much more reasonable to conclude that FTIR was unable to detect any significant change in the LAC values due to solvent extraction. Furthermore, the aromaticity (both AR2 and AR3) of residues after THF extractions is significantly higher than those after the extraction of other solvents (Table 4), which may be related to the fact that THF only dissolves relatively fewer aromatic hydrocarbons from shale samples (Cao et al., 2020), and the residual abundant aromatic hydrocarbons in shale benefits OM aromaticity (Monger, 1984).