FI petrography and microthermometry analyses were performed at the Geological Fluid Laboratory, Xinjiang University, Urumqi, China. Samples were prepared as doubly polished thin sections to a thickness of 0.20–0.25 mm, soaked in acetone for 3–4 h, rinsed with clean water, and dried thoroughly. Twelve samples with abundant and representative FIs were selected for microthermometry. Petrographic observations were carried out on a Nikon LV-150N microscope. Microthermometric measurements were carried out on a Linkam THMS600 heating/freezing stage with a temperature range of -196–600 C. Heating rates were 1–2 C/min approaching phase transitions. Freezing rates were 0.1–0.2 C/min approaching the final ice melting temperature. Precision was ±0.2 C for freezing measurements and ±2 C for heating measurements.

Microthermometric measurements were calibrated using synthetic FI standards for the freezing points of pure CO₂ (-56.6 C) and pure H₂O (0 C), and for the critical point of pure H₂O (374.3 C). Salinities were calculated using the final melting temperature of ice for aqueous two-phase FIs (Bodnar, 1993), and the final melting temperature of NaCl-crystals for halite-bearing FIs (Hall et al., 1988).

Hydrogen, oxygen, and sulfur isotope analyses were performed using a Finnigan MAT-253 mass spectrometer, and lead isotope analyses were performed using an ISOPROBE-T thermal ionization mass spectrometer at the Analytical Laboratory of the Beijing Research Institute of Uranium Geology. Quartz, pyrite, and sphalerite samples were crushed and sieved to 40–60 mesh and handpicked under a binocular microscope, resulting in a purity above 99%.

Oxygen isotopic analyses were performed using the method of Clayton and Mayeda (1963). Oxygen was extracted from the samples by reaction with BrF₅ and converted to CO₂ on a platinum-coated carbon rod for analysis. The hydrogen isotopic composition of FI water was released from the samples by decrepitation of FIs by heating to ∼500 C. The water was then reacted with zinc powder at 410 C for 30 min to generate hydrogen (Friedman, 1953). Results are reported as δD and δ¹⁸O relative to Standard Mean Ocean Water (SMOW) with analytical uncertainties (1σ) of ±2‰ for δD and ±0.2‰ for δ¹⁸O. Samples were reacted with Cu₂O until transformation into pure SO₂ for sulfur isotopic analyses (Robinson and Kusakabe, 1975). Results are reported as δ³⁴S relative to Canyon Diablo Troilite (CDT) with an analytical precision of ±0.2‰. Samples for Pb isotopic analyses were dissolved with concentrated HCl + HNO₃. The two-column AG1-X8 anion resin method was then used to separate and purify the lead. Data obtained were corrected using the ANBS-981 standard for analytical error correction. Analysis accuracy is ±0.005%.

Isotopic data was obtained for three groups (with four quartz samples in each group) representing three mineralization stages. δD_H2O values for stage 1–3 range from −46.5 to −40.7‰, −94.7 to −78.6‰, and −90.3 to −87.8‰, respectively. δ¹⁸O_qz values for stage 1–3 vary from 12.8 to 14.6‰, 16.0–19.2‰, and 11.8–13.8‰, respectively. Since our homogenization temperature data plot within a narrow range for each sample, we selected the mean homogenization temperature value as an estimate of the trapping temperature. δ¹⁸O_H2O values were calculated using the quartz-H₂O equilibrium equation (1000lnα_qz-H2O = 3.38 × 10⁶×T⁻² −3.40) of Clayton et al. (1972). Calculated δ¹⁸O_H2O values range from 2.6 to 4.4‰, 10.3–13.6‰, and 3.7–7.0‰, respectively. Detailed data are listed in Table 2.

Isotopic data was obtained for two groups (with six samples in each group) representing stratiform (stage 1) and veined (stage 3) mineralization, respectively. Stratiform ores record a range of δ³⁴S values from 8.5 to 16.3‰ (average = 13.7 ± 2.9‰, n = 6). With the exception of sample HR-76 (δ³⁴S = 8.5), stratiform samples record a narrow range of δ³⁴S values from 13.1 to 16.3‰ (average = 14.8 ± 1.2‰, n = 5). On the other hand, veined samples yield distinct values from 2.1 to 6.2‰ δ³⁴S (average = 3.4 ± 1.4‰, n = 6), respectively. Stratiform galena samples yield ²⁰⁶Pb/²⁰⁴Pb ratios of 17.206–17.552; ²⁰⁷Pb/²⁰⁴Pb ratios of 15.518–15.523; and ²⁰⁸Pb/²⁰⁴Pb ratios of 37.229–37.503. Veined galena samples yield ²⁰⁶Pb/²⁰⁴Pb ratios of 17.002–17.328; ²⁰⁷Pb/²⁰⁴Pb ratios of 15.502–15.512; and ²⁰⁸Pb/²⁰⁴Pb ratios of 37.025–37.099. Detailed data are listed in Table 3.

Fluid inclusions in stage-1 quartz crystals are simple LV-type FIs. Their homogenization temperatures (206–246 C) and salinities (5.9–11.6 wt% NaCl eq.) are widely variable, and they belong to a low temperature and intermediate salinity NaCl-H₂O system. The proportion of vapor phase is relatively consistent (10–15%), indicating that the ore-forming fluid probably did not undergo mixing or boiling. The limited range of δD_H2O and δ¹⁸O_H2O values for stage 1 samples of the Haerdaban Pb-Zn deposit indicates that quartz may have precipitated from a uniform fluid source. The above characteristics resemble other ancient SEDEX deposits (Samson and Russell, 1987; Wang et al., 2008; Rajabi and Canet, 2012). Seawater with δD_H2O and δ¹⁸O_H2O values close to 0‰ (Ohmoto and Rye, 1979) is considered to be an important fluid source for SEDEX deposits. When plotted on the δD_H2O vs δ¹⁸O_H2O diagram (Figure 6B), isotope data for the Haerdaban deposit fall close to the modified seawater field. Precipitation could result from quenching of metal-bearing submarine hydrothermal solutions in contact with cool seawater, as reported by extensive literature (Sanchez-Espana et al., 2003; Tornos, 2006). The wide range of temperatures and salinities is mainly due to the prolonged period of fluid circulation between submarine hydrothermal solutions and seawater.

Fluid inclusions identified in quartz crystals from stage 2 are also LV-type FIs. Metamorphic fluids are characterized by a NaCl-H₂O system with intermediate temperatures (326–349 C) and low salinities (3.4–6.6 wt% NaCl eq.). Compared to stage 1, FI sizes are smaller, salinities are lower, and the homogenization temperature range is narrower. The difference in fluid nature between the two stages suggests late hydrothermal overprinting. The presence of widespread metamorphic rocks and deformed orebodies and strata gives further evidence of a metamorphic event during mineralization stage 2. Total FI homogenization temperatures were attained in the liquid phase cluster at 326–349 C, which is consistent with lower greenschist facies temperatures (Miu and Ran, 1992). Oxygen isotopes plot well above the field of magmatic water (δ¹⁸O_H2O > 9.5‰). Isotope data plot below the metamorphic fluid field, and trend towards the field of water that interacted with organic matter (Figure 6B). This indicates that the ore-forming fluid in stage 2 was most likely derived from the dehydration of ¹⁸O-enriched metasedimentary rocks during greenschist facies metamorphism and their later interaction with deep-seated carbonaceous slates during fluid migration. The release of δD-depleted organic matter (H₂ and CH₄) from the carbonaceous slate to the metamorphic fluid occurred during regional metamorphism (Ding et al., 2014).

Fluid inclusions identified in quartz crystals from stage 3 are LV-, VL-, and S-type FIs. Vein-type gangue is made up of crystals that formed during hydrothermal overprinting at temperatures of 249–316 C. LV-type FIs are spatially separated from boiling FIAs, indicating they were not trapped synchronously. Boiling FIAs most likely originated from the phase separation of magmatic water due to decompression, while LV-type FIs may represent the input of hot extraneous meteoric fluids. When plotted on the δD_H2O vs δ¹⁸O_H2O diagram (Figure 6B), data fall between the magmatic water field and the meteoric water line. Thus, the ore-forming fluids of stage 3 are hybrid in origin–an initial magmatic source with input of meteoric water. The nature of the fluid and its evolution during stage 3 are similar to those of magmatic-hydrothermal deposits (Li et al., 2018, 2021), which typically consist of vein mineralization controlled by magmatism and faults.

In conclusion, our results suggest that FIs and isotopes from the Haerdaban deposit record three different mineralization events: an early SEDEX event, an intermediate metamorphic remobilization event, and late magmatic-hydrothermal overprinting.

Sulfur is one of the most important ore-forming elements in most metal deposits. Sulfur isotope compositions can trace the source of ore constituents and are widely employed to study deposit genesis (Rye and Ohmoto, 1974). In the Haerdaban deposit, there is a tendency towards δ³⁴S_pyrite> δ³⁴S_sphalerite, which indicates overall equilibrium of sulfur isotopes among sulfides. Two δ³⁴S peaks can be observed in the histogram of sulfur isotope data (Figure 7A). The δ³⁴S values of samples from stratiform mineralization range from 8.5 to 16.3‰ (average = 13.7 ± 2.9‰). Such a wide range of δ³⁴S most likely did not originate from a single source. Except for sample HR-76 (δ³⁴S = 8.5), other samples are enriched in heavy sulfur isotopes, and have limited δ³⁴S values of 13.1–16.3‰ (average = 14.8 ± 1.2‰), which is close to the range of Proterozoic marine sulfates (16‰～17‰, White, 2013) (Figure 7B). The sulfide samples from Haerdaban were mostly formed during a submarine exhalative process. On the other hand, sample HR-76 may have been derived from contamination by interaction with deep magmatic materials. This indicates that both ore-hosting limestone and deep magma reservoirs provided the ore constituents for the stratiform mineralization. The δ³⁴S values of samples from vein mineralization range from 2.1 to 6.2‰ (average = 3.4 ± 1.4‰). Such a limited range indicates a single magmatic source (−3‰～3‰, Ohmoto and Rye, 1979; Hoefs, 2009) (Figure 7B), showing similarities with magmatic-hydrothermal deposits. These sulfide samples precipitated from an epigenetic hydrothermal solution during the magmatic-hydrothermal stage.

Pb isotopic compositions also contribute to understanding the sources of ore constituents (Bierlein and McNaughton, 1998). Pb isotopic data for pyrite and sphalerite samples in the Haerdaban Pb-Zn deposit show values of 17.002–17.552 for ²⁰⁶Pb/²⁰⁴Pb, 15.502–15.523 for ²⁰⁷Pb/²⁰⁴Pb, and 37.099–37.503 for ²⁰⁸Pb/²⁰⁴Pb. Samples plot between the orogenic growth line and the mantle line, or between the orogenic growth line and the lower crust line in the lead isotope compositional diagram (Figure 8). These values indicate that the ore-forming material may have originated from a mixed lower crustal and mantle origin (Stacey, 1975; Doe and Zartman, 1979). The high radiogenic end-member is similar to the crustal lead reservoir (Zartman and Doe, 1981), and is possibly represented by host rocks of the Haerdaban deposit. The low radiogenic end-member is similar to the mantle reservoir, which may correspond to the ore-bearing magmatic fluid. Stratiform mineralization may contain a greater proportion of crustal lead components, since corresponding samples display a higher radiogenic lead ratio (²⁰⁶Pb, ²⁰⁷Pb and ²⁰⁸Pb; Figure 8).

The multistage genetic model is discussed in greater detail below.

Firstly, during stage 1, stratiform and laminated mineralization took place in a SEDEX ore-forming system. Being a syngenetic sedimentary deposit, mineralization and diagenesis occurred almost simultaneously. Thus, the age of SEDEX mineralization in the Haerdaban deposit can be estimated as the formation age of the Haerdaban Group, which was determined as Mesoproterozoic (XBGMR, 1993). Moreover, since the mining district is located on the passive continental margin of the Kazakhstan-Yili Plate (Xue et al., 2014a), its geotectonic location played a crucial role on the occurrence of mineralization. Specifically, upper crustal permeability was enhanced significantly under extensional conditions and resulted in syngenetic faults. These faults provided channels for deep submarine metal-bearing solutions to migrate through, resulting in precipitation by fluid quenching due to contact with cool seawater, which generated the aforementioned stratiform mineralization (Figure 9A).

The ore-forming system subsequently underwent metamorphism and shear deformation in stage 2. The ore and its host rock became folded, and intermediate temperature and low salinity metamorphic fluids were mobilized during this stage (Figure 9B). Stratiform mineralization was remobilized with the assistance of metamorphic fluids, forming sinuous quartz-sulfide veins. Low-grade metamorphism took place at a regional scale, and formed assemblages of biotite quartz schist, quartzite, metamorphic sandstone, and marble. The metamorphic fluids were expelled along deep faults, and deformation resulted in quartz sub-grain formation and grain boundary migration. Structural analysis and chronological research (Wang et al., 2018; Xia and Zhang, 2020) show that the region underwent multistage metamorphism-deformation scenarios. The precise timing of the metamorphic remobilization stage in the mining district is difficult to determine.

Finally, stage 3 was triggered by the intrusion of magma, leading to another enrichment in ore-forming elements, and resulting in mineralization as veins crosscutting or overlying the stratiform mineralization. Veined mineralization may have resulted from the emplacement of diorite. Granitoids developed in the region, yielding zircon U-Pb ages of 467–302 Ma (Hu et al., 1999; Tang et al., 2010; Wang et al., 2012; Huang et al., 2013). Large-scale magmatic activity occurred in a two-peak period; in the Late Ordovician and in the Early Carboniferous. In addition, recent Rb-Sr dating of sphalerite yielded an isochron age of 432 ± 45 Ma (Lv, 2016), which places the stage 3 magmatic event in the Late Ordovician. During this period, the metallogenic environment shifted to continental accretion due to subduction of the North Tianshan Ocean beneath the Yili Plate (Kröner et al., 2012; Xiao et al., 2013; Xue et al., 2014b). A later magmatic-hydrothermal fluid percolated through pore spaces and along faults, and metasomatized the original sediments, resulting in the reactivation and enrichment of ore-forming elements. The pre-existing sulfide minerals recrystallized, purifying the grade of the preexisting orebody. Simultaneously, the addition of meteoric water gradually cooled the magmatic-hydrothermal system, and ore minerals precipitated in structural fissures forming veined mineralization (Figure 9C).

The giant Tekli SEDEX Pb-Zn deposit is located at the southeastern edge of Kazakhstan, close to the Chinese WTO. This deposit has proven reserves of 2,500,000 t Pb and 3,000,000 t Zn, with average grades of 2.8% Pb and 4.19% Zn, respectively (Peng, 1990; Wu et al., 2010). Orebodies are hosted in the Mesoproterozoic Tekli Group of the Riphean Supergroup, and occur as stratiform and lenticular morphologies. The mineralization style of the Tekli deposit is similar to that of the Haerdaban deposit. These two deposits share common features such as the following: 1) ore minerals are predominantly sphalerite and galena; 2) stratiform orebodies occur parallel to host strata; 3) ores display massive and laminated structures; 4) the ore deposits are hosted in Mesoproterozoic sedimentary strata and have a similar geological setting. However, it is worth noting that the Haerdaban mine has evidently undergone epigenetic enrichment and is overlain by later geological processes. A more detailed comparison is shown in Table 4.

The Mesoproterozoic is a SEDEX metallogenetic era of global significance (Leach et al., 2010). Recently, Pb-Zn deposits with SEDEX characteristics have been newly discovered in the Chinese WTO, such as the Haerdaban, Tuokesai, and Sitaihaiquan deposits. These deposits are all hosted in carbon-bearing low-grade metamorphosed clastic rocks, and are located on the passive continental margin of the Kazakhstan-Yili Plate in an extensional setting. Based on previous regional geochemical surveys and remote sensing data, we suggest that the Chinese WTO has great prospecting potential to host a series of SEDEX Zn-Pb deposits.