A total of 55 core samples in the Bashijiqike (K₁ bs) and Baixigai (K₁ b) formations were collected from the BZ9, Bozi-102 (BZ102) and Bozi-301 (BZ301) wells (Figure 1B), with burial depths ranging from 5,836.30 to 7,795.50 m.

Quantitative grain fluorescence (QGF) and quantitative grain fluorescence on extract (QGF-E) can realize the rapid delineation of palaeo-oil zones and current residual oil zones in petroleum reservoirs by detecting the fluorescence attributes of reservoir grains and their extracts after a robust pre-cleaning procedure (Liu and Eadington, 2005). For the purpose of QGF and QGF-E measurement, a Varian Cary-Eclipse fluorescence spectrophotometer from Agilent Technologies was used, which is fitted with a customized sampling stage that allows multiple aliquots to be analyzed automatically.

The core samples were prepared as doubly polished wafers with 80–100 μm in thickness for fluid inclusion analysis. Petrographic observation of fluid inclusions was conducted using a Zeiss Axio Imager A2m microscope. In order to avoid human interference in fluorescence color recognition, fluorescence spectral characteristics of oil inclusions (OIs) and crude oil samples were measured using a Horiba iHR320 imaging spectrometer by coupling with a Zeiss Axio Imager A2m microscope with a UV light source unit. Fluid inclusion microthermometry was conducted using a Linkam THMSG600 heating-freezing stage combined with the cycling technique proposed by Goldstein and Reynolds (1994). The measurement precision is ±1°C for the homogenization temperature (T _h) and ±0.1°C for the final ice melting temperature (T _m), respectively. The heating-freezing stage was calibrated using synthetic CO₂-H₂O inclusion (T _m = −56.6°C) and pure H₂O aqueous inclusion (T _m = 0°C) at low temperature. The NaCl-equivalent salinities of aqueous fluid inclusions were calculated based on an equation proposed by Bodnar (1993).

Vapor volume fractions (φ _vap) of oil inclusions at 20°C were measured using a Zeiss LSM 5 Pascal confocal laser scanning microscopy (CLSM) after three-dimensional volumetric reconstruction. The CLSM delivers a stack of focused, two-dimensional (x–y) fluorescence image slices down through the inclusion (z series). The bright fluorescence of oil within the inclusions with the excitation of 488 nm argon laser light (25 mW), can effectively distinguish it from non-fluorescent gas in the inclusion and the host mineral. Zeiss 3D for LSM image processing software was applied to assist construction of three-dimensional images and volume calculation of oil inclusions.

A LabRAM HR Evolution spectrometer (Horiba Jobin Yvon) equipped with 600 and 1800 g mm^–1 grating was utilized to analyze the compositions of monophasic gas inclusions and diphasic aqueous fluid inclusions. The confocal aperture and slit aperture were set to 1,000 μm and 200 μm, respectively. The excitation wavelength was ∼532 nm (air-cooled, frequency doubled Nd:YAG) with an output power of 300 mW. Before each measurement, the spectrometer was calibrated with the 520.7 cm^–1 Raman peak for silicon. During experiments, the Raman spectra were collected from 100 to 4,000 cm^–1 for 20 s with 10 accumulations. Considering the dominance of CH₄ in the vapor phase of the investigated fluid inclusions, the wavenumber of the stretching vibration band of CH₄ was used to determine the internal pressure of both the monophasic gas inclusions and the vapor bubble of the diphasic aqueous fluid inclusions at room temperature (Lu et al., 2007) after calibration by simultaneously collected Neon lamp signal.

The trapping temperature (T _t) and pressure (P _t) of coexisting petroleum and aqueous inclusions were determined by the double isochore technique (Pironon, 2021). Following the procedure proposed by Aplin et al. (2000), petroleum inclusions were modelled using PVTsim (Calsep, Inc.) software based on the Soave-Redilich-Kwong equation with Peneloux volume correction (Péneloux et al., 1982). The present hydrocarbon compositions in the Tarim Basin were selected as the initial compositions for oil inclusion PVT modelling, which have not been affected by biodegradation, and an iterative analysis was adopted until the calculated vapor volume fraction of an inclusion is consistent with that measured from CLSM at 20°C (Aplin et al., 2000; Munz et al., 2004). The phase envelope and isochore of the petroleum inclusion were then reconstructed.

For monophasic gas inclusions, the P _t can be calculated by combining the internal pressure of the monophasic gas inclusion and the T _t (equal to the T _h of coeval aqueous fluid inclusion in the case of heterogeneous trapping) and equation of state (EOS) for pure CH₄ (Duan et al., 1992). For diphasic aqueous fluid inclusions, combined with the internal pressure of the vapor bubble of the diphasic aqueous fluid inclusions at room temperature, salinity and T _h results and EOS for pure CH₄ (Duan et al., 1992) and CH₄–NaCl–H₂O (Duan and Mao, 2006), the homogenization pressure can be calculated following an iterative approach proposed by Becker et al. (2010). Due to the tiny size of the aqueous fluid inclusions or low CH₄ concentration in the aqueous fluid inclusions, however, the stretching vibration band of CH₄ cannot be detected in most of the aqueous fluid inclusions investigated in the present study. In this case, we assumed the aqueous fluid inclusions as a NaCl-H₂O system and reconstructed the corresponding isochores using the model proposed by Steele-MacInnis et al. (2012).

One dimensional (1D) basin modeling was applied to reconstruct the thermal and burial histories of the BZ9, BZ102 and BZ301 wells using PetroMod 1D software. The input parameters were mainly composed of stratigraphy, burial depth, thickness, lithology, age and erosion. The denudation thicknesses at different tectonic stages in the Bozi area were estimated based on Liu et al. (2018). The boundary conditions were paleo water depth (PWD), sediment water interface temperature (SWIT) and heat flow (HF). The paleo-heat flows were determined according to Lu et al. (2014) and Qiu et al. (2015). The thermal history was calibrated using the present-day measured formation (borehole) temperature data.

The QGF index values obtained at the depths of 7,676.05–7,795.50 m in the BZ9 well range from 3.7 to 11.6, while the QGF-E intensities range from 7.0 to 19.5 photometer counts, with spectral peaks in the range of 356–367 nm (Figure 3). The QGF index values obtained at the depths of 6,758.80–6,866.70 m in the BZ102 well span from 0.4 to 12.9, while the QGF-E intensities are in the range of 6.9–24.4 photometer counts with spectral peaks ranging between 357 and 370 nm (Figure 4). The QGF index values obtained at the depths of 5,836.30–5,930.90 m in the BZ301 well range from 3.4 to 12.3 (Figure 5), which are comparable with that for the BZ9 and BZ102 wells, while the QGF-E intensities fall in the range of 8.1–99.7 photometer counts, with spectral peaks occurring in the ranges of 368–383 nm and 451–460 nm, respectively.

Oil inclusions assemblages (OIAs) in the three wells occur mainly in healed micro-fractures within detrital quartz grains (Figures 6A, B). They are commonly colorless or yellowish under plane-polarized light. Under the UV illumination they show near yellowish-whitish, blue-whitish and bright blue fluorescence colors (Figures 6C, D). OIAs are composed predominantly of diphasic (L_oil-V) and triphasic (S_bit-L_oil-V) oil inclusions with bitumen component occurring along inclusion walls. Oil inclusions primarily have elliptic, elongated or irregular shapes with diameters of 5–10 μm. Two groups of OIAs were recognized based on fluorescence color and bitumen content. The first group (Group I OIAs) are characterized by near yellowish-white and blue-white fluorescing, diphasic (L_oil-V) or triphasic (S_bit-L_oil-V) oil inclusions (Figures 6E, F), with vapor volume fractions (φ _vap) ranging from 2.3% to 7.5% (Figure 7). The second group (Group II OIAs) are typical of bright blue fluorescing, diphasic (L_oil-V) oil inclusions (Figures 6G, H), with φ _vap in the range of 8.2%–11.6% (Figure 7).

Gas inclusion assemblages (GIAs) were also observed in some core samples, occurring mainly in micro-fractures within detrital quartz grains. The GIAs are composed of monophasic gas inclusions with the absence of bitumen component (Figure 6I). Gas inclusions have elongated or irregular shape with 3–5 μm length, and are commonly opaque to semitransparent under plane-polarized light. The gaseous components in the gas inclusions are composed of CH₄ and subordinately CO₂ (Figure 8), with a measured CH₄ Raman spectra shift being approximately 2,910.31 cm^–1.

The micro-fluorescence spectral characteristics of oil inclusions are shown in Figure 9 and Table 1. The wavelength of the maximum intensity (λ _max) of the fluorescence spectra and the ratio of the fluorescence spectral intensity at 650 nm over that at 500 nm (Q _650/500) for Group I OIAs are in the range of 538–552 nm and 0.61–1.37 (average: 0.86), respectively. The λ _max and Q _650/500 values for Group II OIAs fall in the range of 523–532 nm and 0.23–0.72 (average: 0.46), respectively. Three condensate samples collected from the Kuqa Depression have λ _max and Q _650/500 values in the range of 504–511 nm and 0.19–0.27 (average: 0.22), respectively.

A total of 107 fluid inclusions investigated were selected for microthermometric measurements (Figure 10; Table 1). Detailed microthermometric results for the three wells are summarized as follows:

In the BZ9 core samples, a total of 19 oil inclusions and 10 coeval aqueous fluid inclusions were measured for microthermometry (Figure 10A). The T _h values of Group I oil inclusions vary from 48.6°C to 62.8°C, while the coeval aqueous fluid inclusions have T _h values of 104.2ºC–106.5°C. The corresponding T _m values for the aqueous inclusions are in the range from −8.2°C to −7.9°C, with calculated NaCl-equivalent salinities of 15.6–11.9 wt% (average: 11.8 wt%). Group II oil inclusions have T _h values ranging from 16.3°C to 78.5°C, while the coeval aqueous fluid inclusions have T _h values in the range of 114.3ºC–146.2°C. The corresponding T _m values for the aqueous inclusions are in the range from −10.5°C to −9.8°C, with calculated NaCl-equivalent salinity in the range of 13.7–14.5 wt% (average: 14.1 wt%).

In the BZ102 core samples, a total of 33 oil inclusions and 15 coeval aqueous fluid inclusions were measured for microthermometry (Figure 10B). Group I oil inclusions have T _h values ranging from 47.6°C to 66.1°C, while the coeval aqueous fluid inclusions have T _h values in the range of 86.8ºC–101.8°C. The corresponding T _m values vary from −8.2°C to −7.2°C, with calculated NaCl-equivalent salinities of 10.7–11.9 wt% (average: 11.5 wt%). For Group II OIAs, the oil inclusions have T _h values ranging from 39.6°C to 88.5°C, while the coeval aqueous fluid inclusions have T _h values varying between 98.7°C and 130.4°C. The corresponding Tm values range between −10.7°C and −9.8°C, with calculated NaCl-equivalent salinities in the range of 13.7–14.7 wt% (average: 14.3 wt%).

In the BZ301 core samples, a total of 21 oil inclusions and 8 aqueous fluid inclusions were measured for microthermometry (Figure 10C). Group I oil inclusions have T _h values ranging from 41.6°C to 64.2°C, but no microthermometric measurement on the coeval aqueous inclusions was obtained. The bright blue fluorescing Group II oil inclusions have T _h values ranging from 43.8°C to 96.5°C, while the T _h values of coeval aqueous fluid inclusions fall in the range of 107.2ºC–115.8°C. The corresponding T _m values vary from −10.6°C to −10.5°C, with calculated NaCl-equivalent salinities in the range of 14.5–14.6 wt% (average: 14.5 wt%). For the GIAs, the coeval aqueous fluid inclusions have T _h values of 131.6ºC–138.3°C. The corresponding T _m values vary from −12.5°C to −11.4°C, with calculated NaCl-equivalent salinities in the range of 15.4–16.4 wt% (average: 15.8 wt%).

The key input parameters for 1D basin modelling of the BZ9, BZ102 and BZ301 wells are shown in Table 2. Based on the 1D basin modelling results, taking the BZ102 well as an example (Figure 11), it can be seen that from the Late Cretaceous to the Early Paleogene, the strata in the Bozi area underwent a prolonged period of slow subsidence. Since the Late Paleogene, the sedimentation rate of strata has accelerated greatly, especially during the Miocene, and the formation temperature of strata has also increased significantly with depth. There were two important uplifting and denudation events in the study area, one occurred in the Late Cretaceous to Early Eocene (96–65 Ma), resulting in an estimated erosion thickness of 1,100–1,300 m, and the other occurred in the Late Pliocene (3.0–1.8 Ma), causing an estimated erosion thickness of 300–500 m.

The T _h−φ _vap relationship shows that the modelled oils in Group I oil inclusions consist mainly of black oil or volatile oil, whilst those in Group II oil inclusions are primarily composed predominantly of volatile oil (Figure 12). The T _t and P _t of Group I OIAs in the BZ9 well are 128.5°C and 58.3 MPa, and those of Group II OIAs are 141.1°C and 65.0 MPa (Figure 13A; Table 3). The T _t and P _t of Group I and II OIAs in the BZ102 well are 105.6°C and 47.9 MPa and 121.9°C and 58.8 MPa, respectively (Figure 13B). The T _t and P _t of Group II OIAs in the BZ301 well are 131.7°C and 61.3 MPa (Figure 13C). Based on the wavenumber of the stretching vibration band of CH₄ (Figure 8), the internal pressure of the monophasic gas inclusions is 49.2 MPa at room temperature, and the calculated P _t of GIAs in the BZ301 well is 93.0 MPa at 139.1°C (Figure 13C).

Fluorescence color has been commonly used as a thermal maturity indicator for oil trapped in inclusions (Guilhaumou et al., 1990; Stasiuk and Snowdon, 1997; Volk and George, 2019; Ping et al., 2020). With increasing maturity, fluorescence colors would change from yellow to blue (McLimans, 1987). The variations in fluorescence color and fluorescence spectral parameters of oil inclusions in our study indicate an obvious maturity difference of oils charged at these two stages. The near yellowish-white and blue-white fluorescing Group I OIAs represent a relatively “low-maturity” oil compared with that for the bright blue fluorescing Group II OIAs. By combining the fluid inclusion PVT modelling and 1D basin modelling results, the timing of hydrocarbon charge can be determined (Figure 14). The first oil charge event, indicated by Group I OIAs, occurred during the Early Neogene (∼6.5 Ma) in the BZ9 well, while the second oil charge event, as evidenced by Group II OIAs, occurred during the Late Neogene (∼3.5 Ma) (Figure 14A). In the BZ102 well, the first and second oil charge events occurred during the Early Neogene (∼5.5 Ma) and Late Neogene (∼3.0 Ma), respectively (Figure 14B). In the BZ301 well, the timing of the first oil charge event cannot be determined, due to the lack of aqueous fluid inclusions coeval with Group I OIs, and the second oil charge event occurred during the Late Neogene (∼4.4 Ma) (Figure 14C). The timing of gas charge event, as evidenced by GIAs in Figure 6I, is approximately 1.6 Ma. It is worth noting that the reservoir pressure coefficients (ratio of oil inclusion trapping pressure to hydrostatic pressure) were in the range of 1.0–1.3 during the first and second oil charge events (Figure 13), equivalent to normal hydrostatic (formation) pressure or weak overpressure. However, the reservoir pressure reached 93 MPa during the gas charge event, with a calculated pressure coefficient of 1.7 in the BZ301 well (Figure 13C).

During the Late Cretaceous to Early Neogene, the Bozi area experienced a period of stable subsidence (Figure 11A), accompanied by local structural deformation and fault development. With the sedimentation of the Paleogene Kumugeliemu Group thick gypsum salt unit, a complete source-reservoir-cap rock assemblage and overburden strata were in place in the Bozi area (Zhu et al., 2015). During the deposition of the Plicene Kangcun and Kuqa formations, the Bozi area underwent strong compression and structural deformation, causing plastic flow of the Paleogene gypsum salt unit and substantial variations in the thickness of the gypsum salt unit and the Kuqa Formation. Large-scale thrust faults developed simultaneously, even breaching through some thinner sections of the gypsum salt unit (Zhang et al., 2011). This rapid subsidence promoted the Middle Jurassic Qiakemake Formation source rocks to become mature and generate hydrocarbons. These hydrocarbons migrated upward along the thrust faults and charged into the Early Cretaceous traps to form oil pools. A total of two oil charge events have been recorded by OIAs, with the first characterized by near yellowish-white and blue-white fluorescing OIAs, while the second characterized by bright blue fluorescing OIAs (Figure 6). In addition, abnormally high QGF index and QGF-E values were observed at 5,830–5,850 m in the BZ301 well (Figure 5), indicating the accumulation of paleo-oil or residual oil. It is hypothesized that oil was once charged into this sandstone interval with relatively good reservoir petrophysical properties, but the scale of the oil pool was relatively limited, with oil column height of only ∼10 m. The first oil charge occurred during the Early Neogene (6.5–5.5 Ma), while the second oil charge occurred during the Late Neogene (4.4–3.5 Ma) (Figure 14). At this time, the reservoir pressure coefficients were in the range of 1.0–1.3 (Figure 13), indicating the occurrence of a normal hydrostatic (formation) pressure or a weak overpressure in the reservoirs.

Since the Quaternary, large-scale thrust faults in the Bozi area continued to develop and gradually formed the current tectonic configuration (Zhang et al., 2011). This had caused the Jurassic coal or coaly shales to become mature rapidly, entering the stage of dry gas generation (Zhang et al., 2011; Tian et al., 2020). This gas charge event occurred during the Pleistocene (∼1.6 Ma) (Figure 14), and was recorded by the GIAs, which are composed of CH₄ and subordinately CO₂ (Figure 8). The dry gas migrated upward to the Cretaceous traps through faults, and altered the reservoir hydrocarbons via gas flushing, causing the hydrocarbon fluid phase state in the sandstone reservoirs changing from black oil or volatile oil to condensate gas (Figure 15). This is also supported by the changes in fluorescence spectral parameters from oil inclusions to condensate oil samples (Figure 9). The QGF-E spectra peaks in the range of 451–460 nm are indicative of the presence of asphaltene in the reservoirs (Liu and Eadington, 2005), as QGF-E measures adsorbed hydrocarbons extracted from the reservoir grains (Figure 5). This suggests the occurrence of gas-flushing-oil within the reservoirs, which may have caused the exsolution and precipitation of asphaltene from the reservoir oil (Hammami et al., 2000). In addition, the negative deviation of light n-alkanes in the crude oil away from the normal n-alkane profile for normal oil further supports the occurrence of gas flushing in the Kuqa Depression (Zhang et al., 2011). During rapid gas injection, the formation pressure reached 93 MPa with a calculated pressure coefficient of 1.7 in the BZ301 well. Due to the continuous tectonic compression, the formation pressure in the Bozi area continued to increase and eventually became the current ultra-high-pressure condensate gas reservoirs, with the present-day pressure coefficients in the range of 1.8–2.0 (Figure 13).

It can thus be concluded that the presence of excellent sandstone reservoirs, abundant oil and gas sources, efficient cap rocks, and suitable fault activities promoted the formation of Bozi deep and ultra-deep condensate gas reservoirs.

The burial evolution history in the Bozi area is characterized by a prolonged (>100 Ma) shallow burial and a late-stage rapid subsidence (10 Ma-present) (Figure 11). The strata experienced a prolonged period of slow subsidence during the Late Cretaceous to Early Paleogene, which is crucial for maintaining porosity in the sandstone reservoirs. Since the Late Paleogene, the subsidence rate has accelerated abruptly, with an incremental burial depth reaching 2000–3,000 m over a short time span, and consequently a sharp increase of the formation temperature. This unique burial evolution history had promoted the maturation of source rocks and caused an intense hydrocarbon generation, migration, and accumulation (Zhu et al., 2015). In addition, the current geothermal gradient in the Bozi area is approximately 20°C/km, and the formation temperature of the Cretaceous sandstone reservoirs is mostly less than 160°C, lower than the threshold temperature for oil cracking (Waples, 2000; Tian et al., 2006). Under this circumstance, a large quantity of condensate gas can be well preserved. The unique burial history model in the Bozi area is primarily responsible for the stable preservation of liquid hydrocarbons (including condensate gas) in the ultra-deep reservoirs.