The theory that coal-measure source rocks generate gas primarily and oil secondarily has provided a theoretical basis for the rapid development of natural gas industry in China^[1]. In recent years, with the strengthening of deep formation exploration, more and more giant gas fields have been discovered^{[2,3,4,5,6,7]} and most of them are dry gas fields, with little heavy hydrocarbons in the gas^[6,8], indicating the source rocks in these areas have reached high thermal maturity. The Kuqa Depression in the Tarim Basin, where Kela 2, Keshen and other dry gas fields have been discovered, mainly produces natural gas^[9,10,11,12] and provides sufficient gas source for the West-to-East Gas Project. Some in-depth and systematic researches have pushed forward the rapid discovery and efficient development of natural gas in the Kuqa Depression^{[13,14,15,16,17,18,19,20,21]}. However, the low content of heavy hydrocarbons in the above gas fields adds difficulty to the analysis of the gas origin. The Dabei gas field discovered recently contains trace condensate oil which provides important information for determining the origin and source of the natural gas. In this study, hydrocarbon compounds in condensate oil samples were analyzed and identified by high resolution comprehensive two-dimensional gas chromatography-time-of-flight mass spectrometry (GC× GC-TOFMS) to determine the source and origin of the oil and gas by combining with natural gas geochemistry and other analytical data. The study results can provide a basis for understanding the origin and source of natural gas in deep formations and guide oil and gas exploration in deep to ultra-deep formations.

The Dabei gas field is a deep continental giant gas field with very complicated structure^[4]. It is located in Baicheng county of Aksu area, Xinjiang Autonomous Region, and is structurally in the Dabei-Keshen No.5 segment of Keshen zone of Kelasu structural belt in Kuqa Depression, Tarim Basin. It is bounded by the Kelasu fault in the north and Baicheng fault in the south (Fig. 1). Its gas-bearing layer is the sandstone of Cretaceous Bashijiqike Formation, belonging to low porosity and low permeability to low porosity and ultra-low permeability reservoirs, with buried depth of 5 500- 7100 m. It has proven gas reserves of 1 093.19×10⁸ m³, condensate oil reserves of 204.84×10⁴ t, a reserve abundance from (2.86-11.07)×10⁸ m³/km², a gas-oil ratio of 26 357-137 614 m³/m³, a condensate oil content of 11-25 g/m³, and pressure coefficient of 1.54-1.65. It is a large ultra-deep anticlinal gas reservoir with ultra-low porosity, medium-low permeability, high abundance, high output and trace condensate oil^[4].

The well discovered in the Dabei gas field is Well Dabei 1 with a total depth of 6 018 m. It produced 66 431 m³ of gas a day from the 5 568-5 620 m interval of Cretaceous Bashijiqike Formation with 8 mm oil nozzle during formation testing. Gas reservoirs in the fault blocks of the Dabei gas field have independent gas-water contacts, which indicates that the faults there have good lateral sealing ability and separate the (or fault anticline) traps in the fault blocks into independent reservoirs. The Dabei gas field has a geothermal gradient of 2.0 °C/100 m, representing a normal temperature system. It has a pressure gradient of 0.27 MPa/100 m and pressure coefficient of 1.54-1.65, being a high pressure gas reservoir.

The physical property analysis of crude oil in the Dabei gas field shows that the density of condensate oil in surface condition at 20 °C is 0.773-0.815 g/cm³ (Table 1), with an average of 0.801 g/cm³; the dynamic viscosity at 50 °C is 1.130- 2.168 mPa•s, with an average of 1.600 mPa•s; the sulfur content is 0.03%-0.53%, with an average of 0.20%; and the wax content is 6.01%-14.00%, with an average of 8.49%. Generally, it has the characteristics of low density, low viscosity, low sulfur content and condensate is less.

Five condensate oil samples from the Cretaceous Bashijiqike Formation in the DB101-1, DB101-5, DB202, DB207 and DB208 wells were analyzed by GC×GC-TOFMS. The GC×GC system was composed of an Agilent 7890A GC coupled to a hydrogen flame ionization detector (FID) and a liquid-nitrogen-cooled pulse jet modulator. All data were processed with Chroma TOF software.

The one-dimensional chromatographic column was a DB-petro (50 m × 0.2 mm × 0.5 mm). The temperature program used was 0.2 min at 35 °C, increased to 210 °C at a rate of 1.5 °C/min and held for 0.2 min; and increased to 300 °C at the rate of 2 °C/min and held for 20 min. The two-dimen- sional chromatographic column was a DB-17ht (3 m × 0.1 mm × 0.1 μm). The temperature program applied was the same as that for the one-dimensional gas chromatography, but the temperatures were 5 °C higher. The modulator temperature was 45 °C higher than for the one-dimensional gas chromatography. The inlet temperature was 300 °C, the inlet mode was split injection, the split ratio was 700:1, and the sample volume was 0.5 μL. Helium was used as the carrier gas, with a flow rate of 1.5 mL/min. The modulation time was 10 s, 2.5 s of which was the hot pulse time. For the mass spectrometry, the temperatures of the transfer line and the ion source were 300 °C and 240 °C, respectively, the scan range was 40-520 amu, the acquisition rate was 100 spectra/s, and the delay time of the solvent was 9 min.

The GC×GC-TOFMS analysis shows that 2 756, 2 639, 2496, 2 507 and 2 745 compounds were detected in the 5 condensate oil samples at signal-to-noise ratio (S/N) greater than 100. The comprehensive two-dimensional gas chromatography (GC-GC) shows family separation and tile effect, non-homologous compounds appear in different positions due to different boiling points, and homologous compounds show tile-like arrangement due to polarity difference. According to the mass spectrometry information provided by time-of-flight mass spectrometer (TOFMS), the compounds are divided into chain alkanes, cycloalkanes, benzenes, naphthalenes, phenanthrenes, tetracyclic aromatic hydrocarbons, monoadamantanes, diadamantanes, triadamantanes and dibenzothiophenes (Fig. 2a).

The normal alkanes detected in the 5 condensate oil samples are relatively complete. The carbon number of n-alkanes in Well DB208 ranges between nC₈ and nC₃₂, with the main peak at nC₁₅. The carbon number of n-alkanes in Well DB207 is between nC₉ and nC₃₂, with the main peak at nC₁₅. The carbon number of n-alkanes in Well DB101-1 varies between nC₈ and nC₃₃, with the main peak at nC₁₇. The carbon number of n-alkanes in Well DB101-5 falls between nC₈ and nC₃₂, with the main peak at nC₁₆. The carbon number of n-alkanes in Well DB202 is between nC₉ and nC₃₃, with the main peak at nC₁₇ (Fig. 2a). Aromatic hydrocarbons are abundant, not only common benzenes, naphthalenes and phenanthrenes have been found, but also a certain number of tetracyclic aromatic pyrene series have been detected. Because of the small amount of samples, no common biomarkers, such as steroids, hopanes and tricyclic terpenes, were detected. The content of adamantanes was higher, indicating that the samples were all products of high maturity stage. Sulfur compounds are relatively few and only a small number of dibenzothiophenes are found. The 3D plot shows the content and distribution of each compound, in which the volume of peak represents the content of each compound (Fig. 2b).

Compared with the conventional condensate oil, these 5 condensate oil samples have a higher content of aromatic hydrocarbons. Among the aromatics, benzenes, naphthalenes, and phenanthrenes take the first three places in content (Fig. 3a), and a small amount of biphenyls, fluorenes and tetracyclic aromatic hydrocarbons, are also detected. On the comprehensive two-dimensional chromatogram, biphenyls appears slightly later than naphthalenes, fluorenes appears earlier than phenanthrenes and naphthalenes, and tetracyclic aromatic hydrocarbons are marked (Fig. 3a). These aromatic compounds are not obvious in the 3D figure because of their low contents (Fig. 3b).

As a new index for identifying cracking products of high mature oil^{[22,23,24,25,26,27,28]}, adamantane is recognized and affirmed by researchers worldwide, and has been used widely^{[29,30,31,32,33,34,35,36]}. Abundant monoadamantanes and diadamantanes similar in types and contents were detected in all the 5 condensate oil samples (Fig. 4). But only the samples from Wells DB208 and DB207 had a small amount of triadamantane compounds detected.

3.3.1. Monoadamantane series

In TOFMS, adamantane compounds are well separated. On the color contour chroma chromatogram, the peaks of isomers of adamantane with the same carbon number are linearly distributed, and the adamantanes of different carbon numbers show tile effect. The monoadamantane compounds have characteristic ions at m/z =79, 93, 107, and can be selectively separated by different characteristic ions at m/z=136, 135, 149, 163, 177, 191. In the 5 condensate oil samples (Fig. 5a) collected from Wells DB208, DB207, DB101-1, DB101-5 and DB202, 85, 81, 72, 70 and 77 monoadamantanes compounds were detected, with contents of 9 103.40×10^-6, 8 227.92×10^-6, 10 093.09×10^-6, 12542.30×10^-6 and 11 313.13×10^-6, respectively (Table 2). The number of substituent groups in the monoadamantanes ranges from 1 to 5. The monoadamantanes with different numbers of substituent groups show tile-like distribution on the two-dimensional bitmap (color contour chroma chromatogram), and the appearance time for the monoadamantanes with substituents at different positions on the two-dimensional time column is fixed. The names of different adamantanes on the bitmap were determined according to the comprehensive two-dimensional mass spectrogram (Fig. 5a). The 3D plot chromatogram reflects the relative contents of compounds intuitively, the higher the content is, the more obvious the peak, and the darker and flatter the base color will be. Conversely, the light base color and many small peaks are characteristics of impure peaks (Fig. 5b).

As shown in Fig. 6a, nine compounds are labeled with selected ions at m/z=136, 135. According to the mass spectrum information provided by TOFMS, these compounds are classified into 4 types according to the relative molecular mass of 150, 164, 178 and 192, respectively with spacing of 14, and are arranged according to the rule of one '-CH2' difference.

By comparing their chromatograms, it is found that all the nine compounds have the same fragment ions at m/z=79, 93, 107 and the same characteristic ion at m/z=135. According to the "characteristic family separation" on the two-dimensional gas chromatogram, these four types of compounds are alkyl adamantanes with one substituent group in position 1 or 2. According to the substituent positions in methyl adamantane and ethyl-based adamantane on Fig. 6a, a low retention time responds to position 1 and a high retention time to position 2 as inferred by the compounds 2, 3, 4 and 5 with confirmed structural formulas. The difference in relative molecular mass between compound “a” with the relative molecular mass of 178 and the adamantane ion with the relative molecular mass of 135 is 43, and there is no fragment ion distributed between ion peaks of 178 and 135, that is the compound “a” has the difference of a molecular chain with the structure formula of “-CH₂-CH₂-CH₃” with the adamantane, which means that compound “a” is the propyl adamantane with substituent position uncertain, thus, the compound “a” is recorded as C₃-adamantane. As shown in Fig. 6b, 21 compounds are marked by the selected ion at m/z=149, which are classified into four types according to the relative molecular mass of 164, 178, 192 and 206. The difference of the relative molecular mass between compound “b” and the selected ion is 43, which indicates the removal of a molecular chain with the structural formula of “-CH₂-CH₂-CH₃”; and there is a fragment ion peak of 177 between them, suggesting two substituents, one methyl and one propyl; as the substituted position is uncertain, compound “b” is recorded as C4-adamantane. Similarly, the molecular formula of compounds “c”, “d” and “e” can be deduced by this method.3.3.2. Polyadamantane seriesThe diadamantanes in the samples have characteristic ions at m/z=79, 91, 105, and were detected by selected ions at m/z=188, 187, 201, 215, 229. The distribution of diadamantanes is similar to that of monoadamantanes, and the ones with the same number of substituents show tile-like distribution. In the 5 condensate oil samples collected from Wells DB208, DB207, DB101-1, DB101-5 and DB202, 18, 17, 12, 10 and 12 diadamantane compounds were detected respectively (Fig. 7a), with the contents of 781.94×10^-6, 646.31× 10^-6, 556.70×10^-6, 502.50×10^-6, 806.79×10^-6 correspondingly. The number of substituents in them ranges from 1 to 4 (Fig. 7b).

Triadamantanes were detected by characteristic ions at m/z=240, 239, 253, 267. Three triadamantane compounds were detected in the condensate oil sample from Well DB208, with a content of 16.38×10^-6, 2 triadamantane compounds were detected in the condensate oil sample from Well DB207 (Fig. 8a), with a content of 8.01×10^-6, and triadamantane compounds in the condensate oil samples from Wells DB101-1, DB101-5 and DB202 could not be detected because of their low signal-to-noise ratio (Fig. 8b).

Since abundant monoadamantane, diadamantane and triadamantane compounds were detected in the condensate oil samples from the Dabei gas field, and it is generally considered that the formation of adamantanes is related to the crude oil cracking^{[21-25, 32-36]}, which carries high-temperature thermal alteration information of crude oil during reservoir formation, the discovery of these compounds in abundance indicates that the condensate oil in the Dabei gas field is the product of high maturity stage.Different substituted positions of alkyl directly affect the stability of adamantanes. According to the indexes of adamantane maturity^{[35,36,37,38]}, the indexes MAI (MAI=1-methyl monoadamantane/(1-methyl monoadamantane+2-methyl monoadamantane) for monoadamantane and MDI (MDI=4- methyl diadamantane/(1-methyl diadamantane+3-methyl diadamantane+4-methyl diadamantane) for diadamantane were used to calculate the maturity of crude oil. The correspondence between adamantane index and vitrinite reflectance (R_o) is listed in Table 3^[37].

With MAI values of 70%, 73%, 63%, 74% and 67%, MDI values of 49%, 48%, 52%, 54% and 50%, and corresponding R_o ranging from 1.3% to 1.6% (Table 3), the 5 condensate samples from Wells DB208, DB207, DB101-1, DB101-5 and DB202 are all products of high maturity stage.

Natural gas in the Dabei gas field has high methane content and low non-hydrocarbon gas content. Specifically, the natural gas has a methane content of 94.8%-96.5% (95.5% on average), ethane content of 2.11%-2.30%, content of hexane and hydrocarbons heavier of 0.01%-0.36%, nitrogen content of 0.33%-1.27%, and CO₂ content of 0.39%-0.62% (0.51% on average), indicating that the natural gas is the product of high maturity stage of source rock (Table 4).

The condensate oil in the Dabei gas field has a higher tricyclic terpane content, and the relationship between the rearranged hopane content and gammacerane index shows that the condensate oil is high in maturity. The 5 condensate oil samples from the Dabei gas field all have C₂₉Ts/C₂₉H of about 0.4, and C₂₉ sterane 20S/(20S+20R) and ββ/(ββ+αα) of more than 0.4, indicating that the condensate oil is high in maturity (Table 5). The parameters of sterane biomarkers and adamantanes^[15] all indicate that the condensate oil in the Dabei gas field has reached high maturity stage.

The paraffin indexes^[15] of condensate oil from the Dabei gas field, including isoheptane and heptane values also fall in the high mature oil zone, with isoheptane greater than 4 and heptane greater than 20. Among them, isoheptane value = (2- methylhexane + 3-methylhexane)/(cis-1, 3-cyclopentane + trans-1, 3-cyclopentane + cis-1, 2-cyclopentane); heptane value = 100 × heptane/(cyclohexane + 2-methylhexane + 1,1- dimethylcyclopentane + 2,3-dimethylpentane + 3-methylhexane + cis-1, 3-dimethylcyclohexane + trans-1, 3-dimethylcyclopentane + trans-1, 2-dimethylcyclopentane + 3-ethylpentane + 2,2,4-trimethylpentane + n-heptane + methylcyclohexane).

The 5 condensate oil samples from the Dabei gas field have adamantanes content lower than the condensate oil of Kela 2 gas reservoir, equivalent to condensate oil of Well Bozi 1^{[15, 19]}. The condensate oils from Well Bozi1, Daibei area and Kela 2 area differ widely in the content of (3-+4-) methyl diadamantanes, and are 140×10^-6 in Well Bozi 1, 250×10^-6 in Dabei area, and up to 2 000×10^-6 in Kela 2 block respectively. As an index to evaluate the cracking degree of crude oil, (3-+4-) methyl diadamantane has been widely used domestically and internationally^{[30,31,32,33,34,35,36,37,38,39]}. The content of (3-+4-) methyl diadamantane in the condensate oil of Dabei area exceeds 200×10^-6, indicating a high cracking degree of crude oil (>90%).

The natural gas in the Dabei gas field has δ¹³C₁ from -29.3‰ to -33.1‰, and δ¹³C₂ from -20.7‰ to -22.5‰ (Table 4). The δ¹³C₂ of coal-formed gas is generally higher than -29‰, and that of oil-type gas is generally lighter than -29‰^[1]. According to this criterion, the natural gas in the Dabei gas field is typical coal-type gas, which tallies with the humic coal-measure source rocks in the Jurassic-Triassic strata in the Kuqa Depression^{[40,41,42,43,44,45,46,47,48]}. The maturity of natural gas in the Dabei gas field calculated according to the relation between δ¹³C₁ and R_o of coal-formed gas established by Liu Wenhui et al.^[49] is equivalent to source rocks R_o of 1.3%-1.7%, which is consistent with the maturity of condensate oil, indicating that the condensate oil may be formed at the same time with the natural gas. Compared with the Dawanqi area in Kuqa Depression, the natural gas in Dabei area has higher maturity and heavier carbon isotope of methane and ethane, which is related to the larger burial depth and higher thermal evolution of source rocks in this area. As the maturity of source rocks decreases away from the center, gas-oil ratio decreases gradually, and the reservoirs change from condensate gas reservoirs, to condensate gas reservoirs with oil ring, and to oil reservoirs. The diverse phases and states are clearly controlled by thermal evolution stage of organic matter^[44].

The Kuqa Depression have developed Triassic and Jurassic high-quality source rocks, 200-600 m thick, in stable distribution^{[42,43,44,45,46]}. In most part of the depression, the source rocks have reached mature to high mature or over-mature stage, and are 1.5%-3.5% in R_o in the depocenter. The gas generation intensity of the source rocks in the center of Kuqa Depression is above 100×10⁸ m³/km², and gradually decreases away from the center. The gas generation intensity increases linearly in E-W direction, and is maximum in Dina area and Dabei-Keshen area, where the cumulative intensity of gas generation is as high as (350-400) × 10⁸ m³/km². Researches on the subsidence history and thermal evolution history of source rocks in the Kuqa Depression show that the burial depth of Triassic and Jurassic source rocks has been increasing rapidly and the formation temperature increasing with the rapid subsidence of Kuqa Depression since Neogene, and oil generation began about 12 Ma ago. The depositional period of Kangcun Formation (5-12 Ma) was the main oil generation period of the source rocks, when high maturity crude oil and natural gas, and massive condensate oil were generated. The depositional period of Kuqa Formation (2-5 Ma) to the Quaternary period was the mass gas generation stage^[15]. Therefore, the oil and gas captured in the Dabei gas field should be the product of source rocks in the high maturity evolution stage during the depositional period of Kuqa Formation, 5 million years ago, which is of similar origin to oil and gas in Dina 2 condensate gas field^[41]. They are the products of the source rocks entering the generation stage of condensate oil and gas. The condensate oil migrated along faults and accumulated into the Dabei imbricated fault block trap.

The Dabei gas field has good reservoir forming conditions. Affected by the Pliocene tectonic movement, the Dabei area in Kuqa Depression experienced rapid subsidence, which led to a sharp increase in maturity of the source rocks, and the source rocks entered condensate oil generation stage rapidly^{[50,51,52,53]}. The Cretaceous Bashijiqike Formation is located in the advantageous delta front facies belt on the whole, laterally stable in sedimentary facies, and continuous in sand bodies, with thin and discontinuous mudstone, the reservoirs are distributed continuously in large scale^[50]. The overlying thick gypsum-salt layer and its formed overpressure chamber protect the reservoirs from porosity reduction caused by compaction effectively. The strong filling of natural gas has reduced the physical property threshold of reservoir^[51]. The gas charging along with densification, large-scale effective reservoirs and abundant fractures provide storage conditions for the formation of Dabei gas field. The gypsum-salt caprock of Paleogene Kumglimu Formation is thick, and is over 400 m thick except in Wellblock Dabei 1. Bounded by Wellblock Dabei 1, it gradually thickens to more than 1 000 m to the south and north. The gypsum-salt caprock, with large burial depth, high pure salt content, strong plastic fluidity and sealing ability laid a solid foundation for hydrocarbon accumulation in the fault anticline, fault block traps there. The faults provide good migration paths for gas. The subsalt structure of Dabei gas field consists of wedge-shaped fault blocks controlled by basement-involved thrust faults, characterized by large vertical fault throw and dense traps. Under the control of faults, the ascending and descending difference of imbricated anticlines caused juxtaposition of different lithological layers, resulting in lateral sealing and providing trap basis for the formation of Dabei gas field. Clearly, the Dabei gas field has superior geological conditions for gas accumulation and good matching of geological conditions with reservoir-forming factors, conducive to the formation of giant gas field.

The condensate oil samples from 5 wells in the Dabei gas field in the Tarim Basin were analyzed with GC×GC-TOFMS. The results show that the samples have relatively complete n-alkanes, abundant adamantanes, including diadamantanes and triadamantanes. According to calculation, the content of (3-+4-) methyl diadamantanes is more than 200×10^-6. According to the maturity index formula of adamantane, the condensate oil is equivalent to the product of source rock at R_o = 1.3%-1.6%.

The geochemical analysis of natural gas indicates that the natural gas produced from the Dabei gas field is high-maturity coal-formed gas. Based on the relation between carbon isotope composition and maturity, the calculated maturity of the natural gas is equivalent to the product of source rock at R_o = 1.3%-1.7%, suggesting that the natural gas and condensate oil in the Dabei gas field are generated in high maturity stage synchronously.

The hugely thick gypsum-salt rock and gypsum-mudstone in the Paleogene Kumglimu Formation are favorable regional caprocks of the Dabei gas field, and form a good reservoir-cap rock assemblage with the underlying Cretaceous Bashijiqike Formation sandstone. Intensive hydrocarbon generation and sustained strong gas charging of the Triassic and Jurassic coal-measure source rocks, subsalt thrust structures in rows and well developed faults have worked jointly to form the Dabei gas field.