Aqueous inclusions are analyzed in CFC cements, with various sizes (3–50 µm) and irregular shapes. These aqueous inclusions occur as isolated or in trails crosscutting crystals (Figures 5A,C), which are interpreted as the secondary origin. Isolated fluid inclusions are separated from other inclusions by distances much larger than the size of the inclusions, whereas those in trails are relatively close to each other (Figures 5A,B). Two types of aqueous inclusions are distinguished by their phase assemblages and vapor/liquid ratios: (1) liquid-dominated biphase (liquid + vapor) aqueous inclusions, with vapor percentage <50% (L-D) (Figures 5B,D) and (2) liquid-only monophase aqueous inclusions (L-O) (Figures 5B,D). FIAs are distinguished as the most finely recognizable fluid inclusion entrapment events (Goldstein, 2001). The liquid-only and liquid-dominated fluid inclusions commonly coexist within individual trails, and they are considered to belong to the same FIA (Figures 5B,D).

Sixteen groups of FIAs are recognized for microthermometry in CFC. Homogenization temperature (T _h), ice-melting temperatures (T _m-ice), and metastable freezing temperatures (T _f) were measured on 92 L-D biphase inclusions. All the microthermometric data are demonstrated in Table 1, and the results are presented in histograms in Figure 7.

L-O aqueous inclusions were supercooled to −180°C to stretch and induce vapors, and some of them generated vapors once or several times during supercooling. The L-D biphase aqueous inclusions (including the L-O aqueous inclusions with vapor that appeared during supercooling) that coexisted with the L-O aqueous inclusions were tested for T _h values in six FIAs (n = 21). These inclusions exhibit inconsistent T _h values, spreading from 85°C to higher than 200°C (Table 1). The L-O aqueous inclusions were subsequently heated and stretched for T _m-ice measurement. The overall range of T _f values in CFC cements is from −65°C to −40°C, with most values greater than −50°C. These aqueous inclusions have T _m-ice values ranging from −6°C to 0°C (salinities of 0–9.4 wt% NaCl) with bimodal at −0.9°C and 0°C, respectively. Only seven inclusions in this type have T _m-ice values below −1.9°C (Table 1; Figure 7).

Five MCs and 13 CFC samples from the Yingshan Formation in the coverage area were analyzed for carbon, oxygen, and strontium isotopic compositions (Table 2; Figures 8, 9). The δ¹³C values of the MCs have narrow ranges, typically varying from −1.0‰ to −0.4‰ (n = 5). In comparison, the δ¹⁸O values spread from −8.6‰ to −6.4‰, with an average value of −7.5‰. For CFC, the δ¹³C values (−2.3‰ to 0‰, n = 13) overlap with those of the MCs, whereas the δ¹⁸O values (−15.3 to −12.5‰) are notably lower than those of MCs. The ⁸⁷Sr/⁸⁶Sr ratios of the MC samples range from 0.70880 to 0.70894, with an average 0.70888 (n = 5). CFCs have the ⁸⁷Sr/⁸⁶Sr ratios varying between 0.70955 and 0.70990, which are significantly higher than those of the MCs.

The individual LA spot elemental values (n = 11) are summarized in Table 3, and their REEY patterns are illustrated in Figure 6. The total rare earth element (ΣREE) contents of CFC samples vary between 0.65 and 1.60 ppm (mean 0.95 ± 0.30 ppm) (Table 3). The REEY distributions exhibit (1) positive La (1.71–3.42, 2.63 ± 0.61) and Y (2.09–2.48, 2.29 ± 0.14) anomalies, and negative Ce (0.58–0.75, 0.63 ± 0.49) anomalies, (2) superchondritic Y/Ho ratios (55.27–68.92, 60.30 ± 3.73), (3) LREE depletion relative to HREE (Pr/Yb = 0.63 ± 0.15), and (4) MREE enrichment (BSI values of 1.24–1.68, 1.40 ± 0.12).

CFC samples have moderate Sr, Mn, and Fe concentrations of 115.48–263.51, 126.42–176.09, and 386.16–428.60 ppm, respectively (Table 3). The calculated Mn/Sr ratios spread from 0.48 to 1.47 with an average value of 0.76 ± 0.36 (Table 3). All the CFC samples have low Al and Zr concentrations, with Al contents of 0.017–0.318 ppm (n = 9) and Zr contents of 0.011–0.030 ppm (n = 11) (Table 3). The samples exhibit poor correlations among REE, Al, and Zr contents (Figure 10A).

In order to decipher the origin of the diagenetic fluids by aqueous fluid inclusion data, the aqueous inclusion petrography is emphasized here. These measured secondary aqueous inclusions in trails comprise L-O monophase and L-D biphase inclusions (Figure 5). Vapors of the L-O inclusions rarely appear during supercooling, and the coexisted L-D inclusions have inconsistent T _h values (Table 1). The T _f and T _m-ice values of both types of the inclusions mainly distribute in −45°C to 40°C and −1.4°C to 0°C, respectively, suggesting that they have extremely low salinities (Steele-MacInnis et al., 2011; Wilkinson, 2017). Overall, the fluid inclusions in CFC are characterized by inconsistent T _h and low salinities. The findings imply that the L-O aqueous inclusions were trapped in the low-temperature environment (<50°C), and the coexisted L-D inclusions resulted from re-equilibration during burial diagenesis (Goldstein and Reynolds, 1994).

The meteoric water usually has a low salinity with the T _m-ice values close to 0°C in fluid inclusions (Goldstein and Reynolds, 1994; Wilkinson, 2017). The T _m-ice values of −1.4°C to 0°C are higher than modern seawater (−1.9°C) (Figure 7A), indicating that aqueous fluid may trap the paleo-meteoric water. Obviously, the fluid inclusions in CFC are divided into two groups on the T _m-ice histogram, with one group having higher T _m-ice of −0.4°C to 0°C (the highest T _m-ice value is 0°C in individual FIA) (Figure 7B) and the other having relatively lower T _m-ice of −1.4 to 0.7°C (the highest T _m-ice value is −0.7°C in individual FIA) (Figure 7C). These high-salinity inclusions may trap paleo-meteoric waters, which experienced re-equilibration or slightly mixing between freshwater and porewater (Goldstein and Reynolds, 1994; Li et al., 2017).

Most of the MC samples obtained from the Yingshan Formation exhibit that the δ¹³C and δ¹⁸O values are within the respective δ¹³C (−3‰ to +0.5‰) and δ¹⁸O (−10‰ to −6‰) of Early-Middle Ordovician seawater summarized by Veizer et al. (1999). Two samples, however, have lower carbon and oxygen isotopic values (Figure 8). The δ¹³C and δ¹⁸O values of MC samples are positively correlated (R ² = 0.62), indicating that recrystallization occurs in limestones in a closed system, and the porewater revised from seawater dominated in the system (Talbot, 1990; Machel et al., 1996; Arp et al., 2008).

Each δ¹⁸O value of the CFC is particularly lighter compared with the Ordovician marine calcites in the coverage area. On the basis, together with the viewpoint of oxygen isotope fractionation by O’Neil et al. (1969), CFC with light δ¹⁸O value may have been precipitated either from meteoric water or hydrothermal fluid with relatively high temperature.

The ⁸⁷Sr/⁸⁶Sr ratios of calcites are generally interpreted to closely reflect the strontium compositions of the fluids from what they precipitated. The ⁸⁷Sr/⁸⁶Sr ratios of the MC samples are completely within the ⁸⁷Sr/⁸⁶Sr ranges (0.7085–0.7090) of Early-Middle Ordovician marine calcites summarized by Veizer et al. (1999) (Figure 9). In addition, all the CFC samples demonstrate particularly higher ⁸⁷Sr/⁸⁶Sr ratios and Sr concentrations compared with the matrix limestone (Figure 9 and Table 3). These infer that the diagenetic fluid may have obtained more radioactive Sr by interacting with ⁸⁷Sr-rich minerals or detrital rocks.

Consequently, carbon, oxygen, and strontium isotopic compositions of CFC indicate that the parent fluids may be freshwater or hydrothermal fluids. Because the fluid inclusion–trapped freshwaters are of secondary origin and the primary inclusions are masked by secondary ones, it is difficult to differentiate freshwaters and hydrothermal fluids based on C-O-Sr isotopic compositions and fluid inclusions.

The carbonate minerals with low REE concentration are easily contaminated by siliciclastic materials with much higher REE concentrations (Li et al., 2019). Therefore, concentrations of screening elements, such as Al and Zr, were measured to screen possible contaminants in this study. The finding demonstrates that CFC has low Al (mean = 0.209 ± 0.086 ppm) and Zr concentrations (mean 0.020 ± 0.006 ppm). Also, there is noncorrelation among Al, Zr, and ΣREE concentrations (Table 3 and Figure 10A). For these characteristics, CFC is believed to have negligible contamination by siliciclastic material.

The REEY patterns are characterized by LREE depletion relative to HREE, MREE enrichment, high Y/Ho ratios, negative Ce anomaly, and positive La and Y anomalies. MREE enrichment (BSI values >1), Ce enrichment ((Ce/Ce*)_SN ratios approaching 1), and higher Mn/Sr ratios generally resulted from recrystallization during burial diagenesis (Shields and Stille, 2001). The ΣREE is positively correlated to BSI, (Ce/Ce*)_SN and Mn/Sr (R ² = 0.52, 0.63, and 0.56, respectively); that is, the more MREE enrichment, Ce enrichment, and higher Mn/Sr ratios, the higher REE concentrations (Figure 10B). Therefore, CFC calcites may have extremely low REE concentrations (<0.65 ppm), and the recrystallization during subsequently buried diagenesis makes the REE concentrations of CFC higher. The (Ce/Ce*)_SN ratios range from 0.58 to 0.75 (0.63 ± 0.05), and the (Pr/Pr*)_SN ratios are 0.82–1.07 (0.95 ± 0.07) (Figure 10C), suggesting that the CFC samples display an analytically significant positive La anomaly (Bau and Alexander., 2009). The positive correlation between (Ce/Ce*)_SN ratios and BSI (R ² = 0.87) signifies that the decreased negative Ce anomalies may be due to recrystallization during burial diagenesis (Figure 10D).

Moreover, the overall REEY patterns of CFC are similar to the seawater-like proxies, but there still exists differentiation between them. The similarity suggests that the CFC may be inherited from seawater. In addition, REE elements can be largely retained during meteoric diagenesis, with precipitating cements taking up excess REE elements released from dissolved carbonate (Webb et al., 2009). However, depletion in HREE may reflect the removal of HREE by mineral breakdown, alteration, and precipitation during fluid–rock interaction. The REEY results indicate that the CFC may be seawater origin or precipitate from dissolved host carbonate during meteoric diagenesis, but significantly altered by recrystallization. Bau et al. (2003) have argued that the REEY in the diagenetic fluid may be derived from the local country rock.

Because the low-temperature (<50°C) meteoric fluid inclusions are of secondary origin, two scenarios are assumed here: (1) the CFCs are not precipitated from freshwater but are leached by it; (2) the CFCs are precipitated from freshwater and the freshwater are trapped in both primary inclusions, which are not identified by the fluid inclusion petrography and subsequent secondary inclusions. Here, we proposed that the second scenario may be true, and the reasons are as follows.

First, the secondary inclusions suggest that CFC experienced a low-temperature environment, and the oldest time that the samples (the burial depth of 6,100 m in the Well TS3) reached 50°C is approximately the Early Carboniferous, depending on the 1-D burial-thermal history of Well TS3 (Figure 11, Liu et al., 2017). Furthermore, the highest temperature the CFC underwent before the Early Carboniferous is approximately 75°C, and the lowest one is the surface temperature of 20°C. Oxygen isotopic fractionation occurred between calcite and fluid during the precipitation. The parent fluids δ¹⁸O_water (SMOW) values are calculated by the O’Neil et al. (1969) oxygen fractionation equation. With the temperature range (20°C–75°C) and the measured δ¹⁸O values of the CFC samples, the calculated δ¹⁸O_water values of the parent fluids range from −14.4‰ to 2.2‰ (Figure 12). The range of the parent fluids δ¹⁸O_water values and the REEY patterns of CFC both suggest that the diagenetic fluids may be derived from freshwater or Ordovician seawater. However, the Sr isotopic compositions of the calcites are more radioactive than those of the Ordovician seawater (Figure 8 and Figure 9). Therefore, the hypothesis that calcites are precipitated from seawater or porewater that are modified from seawater is excluded.

Second, the REEY patterns of CFC are different from those of the hydrothermal dolomite and calcites (Zhu et al., 2013; Jin et al., 2015; Zhang et al., 2015; Liu et al., 2020). In addition, the CFC has comparatively higher ⁸⁷Sr/⁸⁶Sr ratios and lower δ¹⁸O values in the study area than those of hydrothermal calcites (Figure 8 and Figure 9). Furthermore, any other accompanied hydrothermal minerals are not observed in the study area. These evidences indicate that CFC is unlikely precipitated from hydrothermal fluids.

The REEY compositions are largely controlled by fluid–mineral equilibria when meteoric fluids interact with host rocks because meteoric fluids are virtually devoid of REEY (Debruyne et al., 2016). Thus, CFC can exhibit seawater-like REEY patterns and low ΣREE concentrations (<0.65 ppm), when meteoric water leached marine carbonate (Johannesson et al., 2006; Bourdin et al., 2011), whereas the high ⁸⁷Sr/⁸⁶Sr ratios suggest the fluid leached the overlying sandstone. Bau et al. (2003) observed this characteristic in fluorite from MVT deposits in the Pennine Orefield, England, and suggested that the trace metal irons released from bulk aluminosilicate rocks are incongruent during fluid–rock interaction and fluid migration. Consequently, we propose that the CFC may be precipitated from freshwater that leached the host limestones.

As a quintessential meteoric calcite, CFC was supposed to be precipitated either in the late Middle Ordovician or Devonian periods. It seems unlikely that the origin of CFC could be related to the Late Middle Ordovician event, because the burial depth was less than 600–700 m (Lu et al., 2017), which was insufficient to generate high-amplitude stylolites crosscut by CFC. Zhang et al. (2005) measured the ⁸⁷Sr/⁸⁶Sr ratios of calcites formed during the late Middle Ordovician, which are lower than 0.70940, and corroborated this point simultaneously. In comparison, the ⁸⁷Sr/⁸⁶Sr ratios of CFC precipitated in the Devonian period exceeded 0.70940 (Zhang et al., 2005; Li et al., 2011). Therefore, we inferred that the lighter δ¹⁸O and more radiogenic ⁸⁷Sr/⁸⁶Sr signatures may be mixing of exotic radiogenic ⁸⁷Sr with the detrital rocks. The possible explanation is that Silurian siliciclastic sediments above the Ordovician strata are the likely source for the high radiogenic ⁸⁷Sr in the migrating meteoric water. Besides, pore fluids having interacted with the Silurian sandstones containing feldspathic components could mix with descending meteoric water. On this basis, these CFCs could be ascribed to the meteoric water origin in Early Hercynian karstification during the Devonian period.

According to the above context, the Early Hercynian karstification is responsible for the CFC and well-developed paleocave reservoirs in the coverage area. However, it is worth mentioning that the Yingshan limestone is overlain by the Middle to Upper Ordovician insoluble strata during the Early Hercynian stage, so the limestone cannot be leached directly by freshwater flowing down (Yang et al., 2014). Therefore, based on the geochemical characteristics of CFC samples and paleocave distribution, two possible karst models are as follows: (1) the freshwater flowed down into the Yingshan limestone along with the fracture systems; and (2) the meteoric water migrated laterally along the permeable strata from the buried-hill area.

On the one hand, a large number of strike-slip, tensile, and reverse faults have been formed during the uplift according to previous studies (Lu et al., 2017; Méndez et al., 2019). The faults penetrating insoluble strata provide the main karstification pathways for the initial karstic reservoirs development. In general, meteoric water would infiltrate the Yingshan limestones along the fractures leading directly to the surface under gravity, and the dissolved pores and caves developed ultimately (Figure 13). In addition, the faults and related-karstic reservoirs can also be the hydrocarbon migration passageways and the accumulation traps.

On the other hand, meteoric water can laterally migrate along the highly permeable strata to coverage area from buried-hill area under the paleogeomorphic height difference (Qi and Yun, 2010; Dan et al., 2015). The substratified distribution of paleocaves also supports the possibility of lateral migration of meteoric water. Besides, the farther away from the Upper Ordovician pinch-out boundary, the deeper the paleocaves (Figure 3), which also indicates that the meteoric water may be migrated from the buried-hill area. The dissolved pores and caves develop along the meteoric water lateral migration. During the charging of meteoric water, calcites precipitated in the dissolved pores and caves from the fluid saturated with calcium carbonate. Therefore, calcites precipitated from meteoric water accompanied with depleted δ¹⁸O, high ⁸⁷Sr/⁸⁶Sr ratios, low ΣREE, and low homogenization temperature of fluid inclusion.