Water geochemistry in the Bozhi-Dabei area, Tarim Basin and its implications for natural gas accumulation in deep basin

Abstract

Recently, numerous gas fields have been discovered in deep basins worldwide. However, the geochemistry of the produced water has been poorly studied. Here, we present the chemical and isotopic compositions of water produced from the deep (4–7 km) Bozhi–Dabei (BD) gas field, Tarim Basin, China. The data indicate that the produced water is a mixture of formation water and water condensed from gas. The latter of which has contents of lower total dissolved solids (TDS), different ionic compositions and more negative δD and δ¹⁸O values than that of formation water. The formation water in the Lower Cretaceous sandstone reservoirs consists mainly of meteoric water that was strongly affected by the dissolution of Paleogene halite, forming a regional seal. Meteoric water infiltrated the Kumugeliemu halite during the Kangcun–Kuqa stage (16.3–1.64 Ma, Middle Miocene–Late Pliocene interval). A smaller portion of the formation water consists of iodine-rich evaporated seawater that migrated together with gas from transgression-influenced layers in Triassic–Jurassic source rocks. Using the ¹²⁹I dating model, different episodes of this migration in the Dabei (DB) East District were recognized, although the dating results were not precise. This study enhances our understanding of the migration and accumulation processes of formation water and natural gas in deep basins and suggests that the chemical and isotopic composition of formation water can be used together with that of gas to constrain the sources and accumulation processes of deeply buried gas.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-38393-y.

Introduction

In the field of petroleum geology, one of the most important research objectives is to correlate the sources of hydrocarbon accumulations with potential petroleum sources and elucidate the migration and accumulation processes of the accumulated hydrocarbons. However, previous studies have focused on the hydrocarbons themselves^1–3, and little attention has been given to water, which is also a major geological fluid in petroleum-generating basins. With the continued development of the petroleum industry, interest in the water produced during petroleum recovery has increased^4–6. Numerous components in produced water provide insight into the sources, migration, and accumulation processes of hydrocarbons^7–9; notably, water is involved in many hydrocarbon-related geological events, such as the maturation of source rocks, expulsion and migration of hydrocarbons, and hydrocarbon alteration in reservoirs^10–12. During these processes, some hydrocarbon-related components with specific geochemical characteristics (e.g., iodine, organic acid, and lithium-6) are likely to be distributed in the water as a result of hydrocarbon–water–rock interactions^7,13,14. Among them, iodine (I) is unique. As a biophilic species, it serves as a proxy for hydrocarbon migration, and iodine-129 (¹²⁹I), the only long-lived radioisotope of I, can be used to constrain hydrocarbon accumulation times^15,16. The ¹²⁹I dating approach can provide precise accumulation times for hydrocarbons, times that cannot be obtained through conventional organic geochemical analyses. Although the ²⁹I dating technique has been adopted in numerous case studies, these studies focused on shallow reservoirs (e.g., gas hydrate occurrences and oil accumulations^17–19) rather than on deep basins where gas accumulations are more likely to be found.

With petroleum exploration involving deeper strata, many gas reservoirs have been discovered worldwide^20,21. However, only a few studies involving the geochemistry of the water produced in gas fields have been conducted^22–24. In these studies, investigators sought to distinguish formation water from condensed water, explain the origins and evolution of formation water, and estimate reservoir temperatures using chemical geothermometers. However, the relationships between gas and water have not been explored. Both fluids can migrate together under certain geological conditions. For example, if fault systems link the source rocks and reservoirs and begin opening when petroleum expulsion begins, both the hydrocarbons and pore water in the source rocks may migrate into reservoirs along the fault system. Therefore, the formation water in gas field reservoirs may contain specific hydrocarbon-related information. Detailed geochemical research on the water produced from gas fields would provide additional information on the origin and migration of natural gas.

In this study, a series of water samples were collected from gas-producing wells in the Lower Cretaceous sandstone reservoirs of the Bozhi–Dabei (BD) gas field in the Tarim Basin, northwestern China. The chemical components and several isotopes (e.g., δD, δ¹⁸O, and ¹²⁹I/I) were systematically analyzed. First, formation water was distinguished from condensed water in terms of its chemical and isotopic compositions with the aid of production data. The sources, migration, and evolution processes of the formation water were then revealed. Finally, a dating model based on the geochemical behavior of ¹²⁹I was constructed to reveal the different migration episodes of both natural gas and related water in a specific district. This research could enhance our understanding of the migration and accumulation processes of formation water and natural gas in deep basins.

Geological setting

The Kuqa Depression, which is located in the northern part of the Tarim Basin (Fig. 1a), consists of eight tectonic subunits called the “Five Belts and Three Sags”: the Northern Monoclinal Belt, Kelasu Fold Belt, Yiqikelike Fold Belt, Qiulitage Fold Belt, Southern Gentle Monoclinal Belt, Baicheng Sag, Wushi Sag, and Yangxia Sag (Fig. 1b). This depression is filled mainly with Mesozoic and Cenozoic clastic sediments. Salt sequences are present in the Paleogene Kumugeliemu Formation in the western and middle parts of the Kuqa Depression and in the Neogene Jidike Formation in the eastern part of the depression^25,26. Thus, the Mesozoic and Cenozoic strata are typically divided into three sequences (e.g., suprasalt, intrasalt, and subsalt^27,28; Fig. 1d, Fig. 2). The Paleozoic and Precambrian strata are deeply buried and, thus, are poorly understood.

Fig. 1.

Structural map of the Tarim Basin (a), Kuqa Depression (b) and Bozhi–Dabei (BD) gas field (c) and cross-section along line A–A’ (d, modified from Mo et al.³³). Note: The locations of the Tazhong oilfield and Lunnan oilfield in the platform region of the Paleozoic craton are shown in Panel a. For the purpose of analysis and discussion, the study area was divided into four districts (BZ West District, BZ East District, DB West District and DB East District, Panel c).

Fig. 2.

Simplified stratigraphic column of strata in the BD gas field, Kuqa Depression.

The Kuqa Depression is a superimposed Mesozoic and Cenozoic foreland basin that developed along a passive Paleozoic continental margin²⁹. The evolution of the foreland basin can be divided into three stages: a back-arc foreland basin stage (Triassic), a fault depression–depression basin stage (from the Jurassic to the Paleogene), and a combined rejuvenated foreland basin stage (Neogene to Quaternary)^30,31. The structures in the Kuqa Depression are characterized by a series of thrust belts and fault-related folds. They formed as a result of strong compression and thrusting that were driven by the intracontinental subduction of the Tianshan Orogenic Belt since the Miocene^31,32. These thrust faults act as major pathways for hydrocarbon migration, and fault-related folds are favorable zones for oil and gas accumulation.

The coal and lacustrine mudstones in the Triassic and Jurassic sediments represent two series of major source rocks. The mudstones are composed of type II–III kerogens, whereas the organic matter in the coal consists mainly of type III kerogen³⁴. The Triassic source rocks reached an oil-generating peak at the end of the Paleogene, whereas the Jurassic source rocks began to generate oil during the Miocene^34,35. With rapid burial after the deposition of the Kuqa Formation (< 5.20 Ma), both source rocks entered gas generation stages^34,36. At present, the source rocks in the depositional centers have reached a high and overmature stage (Ro>2.2%)^36,37. The Lower Cretaceous sandstones, including the Bashijiqike (K₁bs) and Baxigai (K₁bx) Formations, are the major gas reservoirs and are dominated by fine- to medium-grained lithic arokoses and feldspathic litharenite, which were deposited in a fan-braided deltaic progradational setting^38–40. The Kumugeliemu salt sequences formed excellent regional seals in the western and middle regions of the Kuqa Depression. They formed in saline lacustrine environments³⁶ but were possibly influenced by sea invasion from the Neo-Tethys Ocean^41,42.

The BD gas field is located to the west of the Kelasu Fold Belt, Kuqa Depression. Gas is the major type of hydrocarbon produced; however, various forms of oil are also found. Most Dabei wells are in the dry gas stage, whereas the wells in the Bozi Block are characterized by wet gas and condensate pools (Table S1). Studies of hydrocarbon-bearing inclusions have shown that the K₁ reservoirs of the east region of the BD gas field contain one form of single-phase CH₄-rich inclusion and two types of oil inclusions with blue–white and yellow–white fluorescence, respectively^43,44. By combining the results of burial and thermal history modeling with homogenization temperature data (Th) of the aqueous inclusions, an oil charging episode was estimated to have occurred at 5–4 Ma, and a gas charging episode was likely to have occurred at 3–2 Ma. Another oil charging episode was not identified because of the inability to obtain Th data for the coeval aqueous inclusions. The yellow–white fluorescence of these oil inclusions, however, indicated that the oil was at a relatively low maturity stage, and this oil charging episode should have taken place at an earlier time (>5 Ma)^43,44. Most oils remigrated after oil accumulation. Some of these oils were lost, while others accumulated again in Neogene–Quaternary reservoirs to form the Dawanqi oilfield (Fig. 1c)⁴⁵.

Samples and methods

All water samples were collected from Lower Cretaceous sandstone reservoirs (K₁bs and K₁bx) at depths of 4,802–7,317 m below sea level (Fig. 1c). Only water samples from gas wells that had been in active production for at least six months were collected; thus, any contamination from drilling additives was avoided. At the sampling site, the collected water samples were filtered through 0.45-μm Millipore filters and collected in clean plastic bottles.

Total dissolved solids (TDS) were measured using the gravimetric method described by Clescerl et al.⁴⁶. The chloride (Cl), bromide (Br) and sulfate (SO₄) concentrations were measured by ion chromatography following appropriate dilution with a Dionex AQUION RFIC instrument, an AS19 ion exchange column and 10 mM KOH eluent. The relative analytical uncertainty was greater than 3.0%. The quantity of iodine (I) was measured using an 8800 inductively coupled plasma–mass spectrometer (ICP–MS; Agilent Corporation, USA). Replicate analyses yielded a precision > 4%. The sodium (Na), potassium (K), calcium (Ca), and magnesium (Mg) concentrations were measured by ion chromatography following appropriate dilution with a Dionex ICS 900 instrument employing a CS12 ion exchange column and 9 mM methanesulfonic acid eluent. The relative analytical uncertainty was greater than 5.0%. Trace elements, including barium (Ba), strontium (Sr), and lithium (Li), were analyzed in diluted solutions using a Varian Vista-Pro ICP optical emission spectrometer (ICP–OES) with an analytical uncertainty greater than 5%.

The δD and δ¹⁸O isotopic compositions were analyzed using a Picarro L2130-I isotopic water analyzer, and the isotopolog concentrations were determined using cavity ringdown spectroscopy. The δD and δ¹⁸O values were recorded relative to those of Vienna standard mean ocean water (V-SMOW). Replicate analyses yielded precisions of ±0.4‰ and ±0.1‰ for δD and δ¹⁸O, respectively. Iodine was separated from water as AgI using an improved solvent extraction procedure^47–49. The ¹²⁹I/I ratios in the AgI precipitates were measured via accelerator mass spectrometry (AMS) using a 3-MV tandem AMS system. The measured ¹²⁹I/¹²⁷I values of the blank samples were approximately 2×10⁻¹³ and were subtracted from those of the actual samples, which were at least three times greater than those of the blanks.

Results

The range of TDS concentrations was 5.59 to 264.10 g/L (Table S1). On the basis of the definition of water salinity^5,50, the water samples were classified as brine (>35 g/L; 16 samples), saltwater (10–35 g/L; 12 samples), or brackish water (1–10 g/L; 5 samples). The water obtained from the DB East District was generally more saline than the water from the other districts were. In most water samples, Cl and Na were the predominant anion and cation, respectively. However, SO₄ was the predominant anion in three water samples (BZ12, BZ1201, and BZ1202), with concentrations ranging from 3,637.80 to 8,183.00 mg/L. The chemical compositions of these water samples were broadly consistent with previously published data for oilfield water^4–6.

As shown in the Piper diagram (Fig. 3a), most water samples were grouped in the Cl and Na+K corners and were classified as Na–Cl-type water. The exceptions were BZ12 and BZ1201 (Na-SO₄ type). Consistent with the Piper diagram, most water samples were dominated by Cl and Na in the Schoeller diagram, with BZ12 and BZ1201 as exceptions (Fig. 3b). Interestingly, most of the water samples (in either the Piper or Schoeller plots) were characterized as close to seawater; that is, some were similar to the brine in the deep reservoirs of the platform region of the Paleozoic craton, Tarim Basin^51,52, although they were collected from deltaic sandstone reservoirs. In addition, the evident disparity between water samples obtained from local surface water^53,54 indicated that these produced water samples were not contaminated by drilling additives, which are commonly mixed with local surface water during drilling operations.

Fig. 3.

Piper (a) and Schoeller (b) diagrams of the produced water in the BD gas field.

The δD values varied between -94.82‰ and -53.76‰, and the δ¹⁸O values were mostly positive, ranging from -5.83‰ to 2.98‰. Both values are very different from those of seawater (0‰ and 0‰) or local surface water (-82% to -70% and -12.2% to -10%)^53,54, which suggests a mixed origin for the water and possibly strong water–rock interactions. The ¹²⁹I/I ratios in all the water samples varied between 296 and 4,870 ×10^-15. These ratios are significantly lower than those in the surface environment after the nuclear era (typically > 10,000 × 10^-15₎[7], thus suggesting that anthropogenic ¹²⁹I inputs in these water samples are negligible and that contamination during the sampling protocols was prevented.

Discussion

Influence of condensed water

Condensed water accounts for a certain proportion of the produced water in gas reservoirs—especially in reservoirs with high gas–water ratios (GWRs)—and is thus an important factor in determining the origin of formation water, which is in the liquid phase in reservoirs^22,24,55. Accordingly, we sought to estimate the influence of condensed water on the produced water that was obtained from the study wells. The TDS contents of the produced water were negatively correlated with the GWRs (Fig. 4a). The produced water samples that were associated with high gas yields had low TDS values. Lico et al.⁵⁶ suggested that water produced at gas wells with GWRs <30,000 m³/m³ is not diluted by condensed water. On the basis of these criteria, only six water samples in the DB East District (DB101-1, DB101-5, DB102, DB208, DB209, and DB304) were classified as formation water; the other water samples were diluted by condensed water to varying degrees. As evidenced by the well production logs (Fig. 4b), the TDS levels in the produced water in the early stage of gas well production were relatively low but increased with continued gas production. These findings are consistent with the results of previous studies, which suggest that the invasion of formation water with increasing gas production could decrease the GWR and thus reduce the proportion of condensed water in the produced water^22,24.

Fig. 4.

Relationships between the gas-to-water ratio (GWR, m³/m³) and total dissolved solids (TDS, g/L) in the sampling wells in this study (a) and from well production logs (b) from the BD gas field.

Condensed water is considered to have low salinities and relatively low isotopic compositions^22,57. In the BD gas field, five samples of brackish water (DB1202, DB11, DB17, BZ1JS, and BZ12) had significantly lower TDS (5.59 to 9.72 g/L) and negative δD (-75.08 to -65.22‰) and δ¹⁸O (-4.06 to -1.62‰) values than the other samples did (96.94 to 264.10 g/L of TDS, -59.82 to -53.76‰ of δD, 0.64 to -2.98‰ of δ¹⁸O). The isotopic compositions of condensed water were controlled mainly by temperature-dependent equilibration fractionation. According to the equilibration fractionation equation of Horita & Wesolowski⁵⁸, the condensed water that equilibrated with the formation water in reservoirs had theoretical δD and δ¹⁸O values (-78.11‰ to -72.00‰ and -2.81‰ to -1.13‰, respectively), similar to the measured values for one of the brackish water samples (BZ12, which had the lowest δD value of all the brackish water samples). As such, the five brackish water samples from the BD gas field can be considered condensed water. In terms of ionic compositions, these condensed water samples were relatively enriched in HCO₃ and SO₄ and depleted in Cl, and the ranges of K, Na, Ca, and Mg were more variable than those in the formation water samples were (Fig. 5a and 5c). The reason for this ionic composition was not clear. The ionic compositions of condensed water are controlled by numerous geological and gas-production-related processes (e.g., ion partitioning between vapor and liquid water, dissolution from adjacent minerals, and contact behavior among water, hydrocarbons, and rocks)⁵⁵. However, low salinity has always been considered a typical characteristic of condensed water.

Fig. 5.

Major ionic compositions of the formation water (a), mixed water (b) and condensed water (c) in the BD gas field.

The ionic compositions of the formation water and condensed water differed from those of the local surface water (Fig. 5a and 5c); the formation water was very similar to the brine present in the deep reservoirs of the platform region of the Paleozoic craton, Tarim Basin, except for the low content of Ca in the formation water (Fig. 5a). Furthermore, other water samples exhibited mixed compositional characteristics, such as combined formation water and condensed water, and thus were considered mixed water (Fig. 5b).

Two endmembers of formation water

The water samples from the BD gas field are distributed on the right side of the global meteoric water line (GMWL) and far below the seawater and seawater evaporation trajectory (SET; Fig. 6). The formation water plotted in the same region as the water from the platform regions in the same basin, implying that they shared a similar origin. The water in the platform regions formed via the mixing of meteoric water and evaporated seawater, followed by a positive δ¹⁸O shift during long-term water–rock interactions^51,52. As such, the mixture of evaporated seawater and meteoric water could be a source of the formation water in the basin. The mixed water plotted in a region located between those where the formation water and local surface water (also local meteoric water) plotted. The condensed water plotted in the middle of the data cluster, and compared with the condensed water, some of the mixed water had lower δD and δ¹⁸O values. It appears that another recent meteoric water influx likely occurred in the mixed water reservoir. If this was the case, recent meteoric water would have dissolved the E_1-2km halite as the water infiltrated downward and thus resulted in high salinity. However, the low TDS values of the mixed water contradict this explanation. Furthermore, the δD and δ¹⁸O values of condensed water under geological conditions could be lower than the values that were calculated from the equilibration fractionation equation by Horita & Wesolowski⁵⁸ because the isotopic fractionation process is controlled not only by temperature but also by other factors (e.g., salinity, pressure, and isotopic exchange with other oxygen-containing and hydrogen-containing compounds). Therefore, the lower levels of δD and δ¹⁸O isotopes in mixed water can be attributed to mixing with condensed water rather than a recent invasion of meteoric water.

Fig. 6.

Plot of δD vs. δ¹⁸O values of the produced water of the BD gas field.

Compared with seawater, the produced water had lower Br contents (1.00–42.31 mg/L) and higher Cl/Br mass ratios (1,719–16,320), as shown to the left of the SET in Fig. 7a. Among all the different types of water, formation water with high Cl and Br contents plotted in a region with halite-dissolution water. These findings indicate that the Cl and Br in the formation water were derived mainly from halite dissolution. Halite that is deposited in sedimentary basins typically has low Br concentrations (50–200 ppm) because of the low partitioning coefficient of Br into mineral lattices during halite precipitation^6,59. Consequently, the dissolution of halite yields water with high Cl/Br ratios. However, halite-dissolution water is characterized by variable Br contents (3.5–97 ppm) and high Cl/Br ratios (1,246–43,311 ppm)^60–62 because halogen partitioning behavior during halite precipitation and dissolution is controlled by many geological factors (e.g., source of solutes, stage of halite precipitation, and water–rock ratio when dissolution occurs)⁶. In the BD gas field, the endmember of evaporated seawater is characterized on the basis of the δD–δ¹⁸O relationship. Therefore, the formation water is mixed with a high proportion of halite-dissolution meteoric water and a low proportion of evaporated seawater. Furthermore, the TDS values (96.94–264.10 g/L), which were lower than the solubility of sodium chloride in water (360 g/L at 25 °C), demonstrated that the formation water was not saturated with halite and that another type of water (evaporated seawater) contributed.

Fig. 7.

Plots of Cl vs. Br (a) and I vs. Br (b) for the produced water from the BD gas field. The seawater evaporation trajectory (SET) in both figures was generated using data from Fontes and Matray⁶³ and Zherebtsova and Volkova⁶⁴. “Dilution” indicates the dilution of seawater by freshwater. “GP” and “HP” indicate the stages in which gypsum and halite start to precipitate, respectively. Halite-dissolution water data were obtained from Kloppman et al.⁶⁰, Pinti et al.⁶¹ and Worden et al.⁶². Other symbols are defined in Fig. 5.

The approximate proportions of halite-dissolution meteoric water and evaporated seawater were also calculated. If the δD values of local meteoric water (-80‰) and seawater (0‰) were selected as the endmembers, the contributions of meteoric water to formation water were between 67.2% and 74.7%. If the lowest content of Br in halite-dissolution water (4 mg/L⁶¹) and the Br content in evaporated seawater at the stage where gypsum began to precipitate (234 mg/L⁶³) were selected as the endmembers, the ionic contribution of halite dissolution in the formation water ranged between 87.4% and 95.1%. Therefore, the major contribution of halite-dissolution meteoric water to the formation water was confirmed.

Origin of formation water

The iodine contents of the water samples (0.77–16.12 mg/L) were higher than those of the seawater and the SET (0.06 mg/L⁶⁴); accordingly, the samples are plotted in the upper left zone of the SET in Fig. 7b. In sedimentary basins, the enrichment of I in water is not due to any kind of water‒rock interaction because the large ionic radius of I makes it difficult to incorporate into mineral lattices and instead is solely attributed to the degradation of organic material during source rock maturation because of the biophilic nature of I^6,65. Therefore, such high amounts of iodine were not sourced from halite-dissolution meteoric water. However, it is unclear when I was released into the pore water in the source rock. Since the generated oils are universally depleted in I (0–23.1 mg/g oil^66–68) in comparison to those from organic matter-rich (OM-rich) shales (195–6150 mg/g rock⁶⁹), it is highly likely that the pore water from the source rock that was obtained after the oil window could already be rich in I. Thus, the seawater evaporated from the deep strata could be expelled from the mature source rocks or pass through the mature source rocks and thus participate in I uptake.

Although the Triassic–Jurassic source rocks were deposited in swamp and lacustrine settings, many OM-rich layers, including Qiakemake (J₂q), the upper layer of Yangxia (J₁y), and Huangshanjie (T₃h), have been demonstrated to have experienced Tethys Ocean invasion, and the quality of organic matter greatly increased in those instances^70,71. Furthermore, both the oil–source correlation and the gas–source correlation in previous studies indicate that both hydrocarbon types in the BD gas field are humic organic matter and thus are confidently derived from Triassic–Jurassic source rocks^72–74. As such, the evaporated seawater can be derived from OM-rich Triassic–Jurassic strata, which were influenced by seawater invasion. Paleozoic marine strata are another possible source of evaporated seawater. However, related geological information is generally limited because these strata are deeply buried at depths > 8,000 m, and no drilling operations in the study area report the occurrence of Paleozoic marine strata. Although some Carboniferous sediments have been discovered in outcrops near the Tianshan Orogenic Belt, all are organic-lean rocks²⁵. Furthermore, the inclusion analysis of the K₁ sandstone reservoirs indicates that the homogenization temperature of aqueous inclusions was never higher than 150 °C, which further indicates that hot fluids from Paleozoic marine strata did not occur^43,45. In addition, the previous oil/gas correlation also excludes any fluid migration from Paleozoic marine source rocks, which should take plankton and bacteria as major biological inputs^72–74. Therefore, on the basis of the present geological information, any fluid migration from Paleozoic strata is difficult to verify. Overall, the OM-rich layers that were influenced by seawater invasion in the Triassic–Jurassic strata were the most likely sources of I-rich evaporated seawater.

The evaporated seawater occupying the pores of OM-rich layers could take I up from kerogen and then migrate upward with natural gases during the maturation of source rocks. Previous studies have demonstrated that OM-rich shale can contain a certain amount of connate water (water saturation > 10%) after compaction during the diagenetic stage^75,76. These water molecules are stored in organic and inorganic pores with diverse pore widths, mainly through monolayer adsorption⁷⁷. However, the water content evolves with both hydrocarbon maturation and mineral transformation. Thermal simulation of a selected shale core sample revealed that the water content in this OM-rich shale varied greatly (9.77–32.20 mg/g rock)⁷⁸. The increase in water content during the thermal simulation was attributed to the release of water from clay transformation, including the illitization of smectite and degradation of kaolinite, whereas two reductions in water content during the maturation process indicated two coexpulsion events for both pore water and hydrocarbons (pore water plus oil and pore water plus natural gas, respectively). The displacement of liquid and gaseous hydrocarbons is the major driving force for water expulsion. In terms of the BD gas field, a certain amount of evaporated water as connate water could be retained in the pores of OM-rich layers that were influenced by seawater invasion and then take I up from kerogen during the maturation process. Afterward, with the further maturation process, some of the I-rich evaporated water was expelled with generated crude oils, whereas the others migrated outside the source rock with natural gases during the late period. The water released from clay transformation was suggested to be fresh^79,80, but its ionic composition was not clear. The high TDS content of the formation water of the BD gas field suggested that the contribution of water released from clay transformation might be very limited.

In the BD gas field, the Kumugeliemu salt sequence (E_1-2km) contains several layers of gypsum-bearing mudstone, gypsum, and halite, with a total salt thickness of 100–3,000 m. This sequence was deposited in a lagoon environment but was influenced by invasion from the Neo-Tethys Ocean, which flooded the Tarim Basin via the Awati Strait to the west^41,42. The meteoric water would have infiltrated downward—dissolving halite after the E_1-2km salt sequences were deposited. However, the evaporated seawater, as another endmember of the formation water, could not be derived from the residual seawater if it was trapped in the E_1-2km halite. Such a mixture of residual seawater and halite-dissolution meteoric water could not explain the enrichment of I in the formation water. As such, the residual seawater in E_1-2km could be displaced after halite precipitation and thus did not contribute to the formation of water. The deep source of the evaporated seawater is again confirmed. The connate water in the reservoirs was of meteoric origin, as evidenced by the fact that the Lower Cretaceous sandstone was deposited in a fan-braided deltaic progradational setting, but the water in this environment might have been saline or brackish because of evaporation in a tropical and arid paleoclimate^38,40. However, the connate water should have been displaced by meteoric water during the denudation period in the Late Cretaceous. Here, the high TDS content of formation water indicated that both connate water and meteoric water with fresh characteristics were highly likely to be displaced by later geological fluids and thus did not remain. Therefore, the formation water was likely mixed with large proportions of meteoric water that infiltrated the E_1-2km halite, and a small proportion of I-rich evaporated seawater migrated upward from OM-rich Triassic–Jurassic source rocks. During gas production, formation water was mixed with condensed water in most wells—to varying degrees—to form mixed-source water.

Evolution process of formation water

The evolution process of the formation water was strongly related to the maturation history of the Triassic–Jurassic source rocks and local tectonic events. Six source layers in the extensive Triassic and Jurassic strata were demonstrated to sequentially generate oil from the end of the Paleogene to the end of the Miocene (65.5–5.20 Ma)^34,35. At nearly the same time, with compression by the Tianshan Orogenic Belt, large-scale thrust faults began to develop from north to south in the Kuqa Depression, and some low-amplitude anticlines developed on the tops of the thrusts⁷³. The thrust faults linked these source layers to traps and thus provided effective migration pathways for the expelled oil. Consequently, oil accumulated in the K₁ sandstone reservoirs, with the E_1-2km salt as the seal. During the Middle Miocene to the Pliocene (Kangcun–Kuqa stage, 16.3–1.64 Ma), thrust fault systems developed further, and the E_1-2km salt began to experience plastic flow. Some thrust faults penetrate thin sections of the E_1-2km salt layer. At this time, oil charging continued, but most of the accumulated oil migrated upward through the E_1-2km salt. From the Pliocene to the present (< 5.20 Ma), with the strengthening of thrust faults, a series of imbricated fault-block traps developed. The source rocks rapidly generated natural gas under the massive deposition of Kuqa and Quaternary strata (Pliocene to present, < 5.20 Ma). These gases migrated upward and accumulated in subsalt sandstone reservoirs to form the present DB gas field^45,73.

An endmember of formation water, namely, halite-dissolution meteoric water, likely intruded into reservoirs during the Middle Miocene to the Pliocene (Kangcun–Kuqa stage, 16.3–1.64 Ma) when the accumulated oil in the K₁ sandstone reservoirs was altered. The thrust faults penetrating the E_1-2km salt layer provided the opportunity for the halite-dissolution meteoric water to move downward, a mechanism similar to that of the upward-migrating crude oils. Another endmember, namely, evaporated seawater, may have migrated together with natural gas from the source rocks rather than with crude oil. As mentioned above, the early-charged oils in the K₁ reservoirs were altered from the Middle Miocene to the Pliocene (Kangcun–Kuqa stage, 16.3–1.64 Ma), after which the reservoirs were filled with natural gas from the Pliocene to the present (< 5.20 Ma). Under these conditions, brine that migrated with crude oil could be largely displaced by late-charging fluids (e.g., halite-dissolution meteoric water migrating downward and water migrating upward with natural gas). Research on fluid inclusions has demonstrated that aqueous inclusions associated with late-charging gas (TDS; 150–250 g/L) are more saline than aqueous inclusions associated with early-charged oils (TDS<127 g/L)^43,44. The former values are highly consistent with the measured salinity of the formation water in this study (TDS; 218.31–264.10 g/L, with the exception of 96.94 g/L). Therefore, the comigration of evaporated seawater and natural gas was confirmed.

The chemical composition of the formation water was diagenetically altered to a large degree. In addition to the halite dissolution mentioned above, the interactions were dominated by those with different kinds of clastic minerals; thus, the enrichment of Li and K and, partly, the enrichment of Ca in the formation water occurred (see Supplementary materials, Fig. S1). However, because the clastic sediments were distributed in all the Mesozoic–Cenozoic strata in the BD gas field, it is not clear in which strata these interactions occurred.

Dating model to constrain the migration times of evaporated water in the DB East District

Iodine-129 (¹²⁹I, t_1/2=15.7 Myr) is the only long-lived I radioisotope. Under natural conditions, ¹²⁹I can be generated by the spallation of Xe isotopes in the atmosphere (cosmogenic mode) and the spontaneous fission of ²³⁸U in the crust (fissiogenic mode)⁷. Cosmogenic ¹²⁹I is homogeneous in all surface reservoirs, with an initial prenuclear equilibrium ¹²⁹I/I ratio of 1,500 × 10^{-15 81}. When a fluid is isolated from the surface environment, ¹²⁹I decays as follows:

1

where R is the ¹²⁹I/I ratio of the sample, R_i is the initial ¹²⁹I/I ratio (1500×10⁻¹⁵), λ₁₂₉ is the decay constant of ¹²⁹I (4.41×10⁻⁸ yr⁻¹), and t is the time elapsed since isolation.

During burial, the proportions of fissiogenic ¹²⁹I in the fluids increased. The abundance of fissiogenic ¹²⁹I was calculated using the following equation⁸²:

2

where N₁₂₉ is the number of ¹²⁹I atoms in a fluid sample, N₂₃₈ is the number of ²³⁸U atoms in a rock sample, λ₂₃₈ is the decay constant for the spontaneous fission of ²³⁸U (8.5×10⁻¹⁷ yr⁻¹), λ₁₂₉ is the decay constant of ¹²⁹I (4.41×10⁻⁸ yr⁻¹), Y₁₂₉ is the production rate for mass 129 (0.0003), ρ is the rock density, E is the proportion of fissiogenic ¹²⁹I released from the rock, P is the effective porosity of the rock, and t is the time that the fluid has been in contact with the rock.

Because the dilution effects of condensed water in mixed water were not quantitatively evaluated, the assessment of ¹²⁹I data focused on the water samples in the DB East District, encompassing all the formation water samples and two other mixed water samples (DB304-1 and DB202). DB304-1 and DB202 might be influenced by dilution water to some degree because of their higher TDS contents and greater quantities of positive δD-δ¹⁸O isotopes than those of other mixed water. The ¹²⁹I concentrations in the DB East District ranged between 12.42 and 42.90 atoms/μL and increased with a roughly west-to-east trend (Fig. 8). The ¹²⁹I concentrations were higher than those in prenuclear meteoric water (approximately 0.07 atom/μL⁸³). This indicates another contribution from fissiogenic ¹²⁹I from rocks.

Fig. 8.

¹²⁹I distributions in water samples from the BD East District. The sample symbols are defined in Fig. 5.

These water samples are mixed with deep evaporated seawater and halite-dissolution meteoric water. Halite-dissolution meteoric water received cosmogenic ¹²⁹I before moving downward. Evaporated seawater can obtain fissiogenic ¹²⁹I from U-enriched source rocks. ¹²⁹I from both meteoric water and evaporated seawater could have been delayed after both fluids entered the reservoirs. However, both meteoric water and evaporated seawater could obtain fissiogenic ¹²⁹I from the surrounding sandstone reservoirs. In conclusion, the ¹²⁹I concentrations in the water from the DB East District were composed of 1) cosmogenic ¹²⁹I in meteoric water, 2) fissiogenic ¹²⁹I from sandstone reservoirs in meteoric water, 3) fissiogenic ¹²⁹I from source rocks in evaporated seawater, and 4) fissiogenic ¹²⁹I from reservoirs in evaporated seawater. These relationships can be expressed using the following equation in four parts:

3

where C₁₂₉ is the measured ¹²⁹I concentration in water; x is the proportion of evaporated seawater; N_MW129 is the number of ¹²⁹I atoms in prenuclear meteoric water in an equilibrium state; t_MW and t_S are the times the meteoric water and evaporated seawater entered the reservoir, respectively; N_R238 and N_S238 are the numbers of ²³⁸U atoms in the reservoir and source rocks, respectively; ρ_R and ρ_S are the densities of the reservoir and source rocks, respectively; E_R and E_S are the proportions of fissiogenic ¹²⁹I released from the reservoir and source rocks, respectively; P_R and P_S are the effective porosities of the reservoir and source rocks, respectively; and t_eq is the time the evaporated seawater contacted the source rocks before being expelled.

In Equation (3), t_S is an unresolved parameter. To calculate t_S, the following data were used: x was calculated using the δD value of the formation water, and -80‰ and 0‰ were assumed to be the endmembers of meteoric water and evaporated seawater, respectively. The calculated x values varied between 25.0% and 30.2% (Table S2). N_MW129 was 0.074 atoms/μL, as obtained by multiplying the number of I atoms in meteoric water (4.95×10¹⁰ atoms/μL⁸⁴) and R_i (1500×10⁻¹⁵⁷⁴). The t_MW values ranged between 16.3 and 1.64 Ma according to the above discussion. N_R238 was 4.02×10¹⁸ atom/kg, which was calculated from the average U content in the K₁bs sandstone (1.59 mg/kg; n=9⁸⁵). Because the shales and coals in the Triassic–Jurassic strata have been suggested to generate natural gas in the BD gas field^73,86, the average U content of both types of source rocks (4.37 mg/kg; n=35; measured in this study; Table S3) was used to calculate N_S238 (1.11×10¹⁹ atom/kg). ρ_R is 2.60 g/cm³ for sandstone⁴⁸. The ρ_S values should vary between 1.50 and 2.60 g/cm³, which are the average densities of anthracite (high-rank coal)⁸⁷ and black shale¹⁵, respectively. The E/P ratio was used to indicate the escape probability of ¹²⁹I from the rock into fluids. This parameter cannot be obtained precisely but was commonly assumed to be in the range of 1–30 in previous studies⁷. It is believed that ¹²⁹I can escape from source rocks into pore water rather easily because compared with U in mineral lattices in organic-poor sediments, U adsorbed on organic material in source rocks more readily releases ¹²⁹I from U fission to pore water^52,88. As such, E_R/P_R is assumed to be 3 for sandstone reservoirs, whereas E_S/P_S is assumed to be 15 for source rocks. The value for t_eq is 90 Ma or any time prior.

Because both t_MW and ρ_S had two endmember values, the calculated t_S value of the evaporated seawater from each well varied within a certain range (Table S2, Fig. 9a). However, the results demonstrated that the evaporated seawater migrated into the reservoirs at different times. The evaporated seawater in DB304 and DB304-1 was the youngest (t_S-avg=5.70 and 12.41 Ma), whereas that in DB101-5 was the oldest (t_S-avg=68.97 Ma). The t_S-avg values for the other four water bodies ranged from 22.73 to 36.31 Ma. Overall, the charging times for the evaporated seawater decreased eastward.

Fig. 9.

Calculated migration times of evaporated seawater (t_S), with a west-east-trending distribution (a), and the relationship between t_S and the dryness coefficient of natural gas (b) in the BD East District.

Different episodes of evaporated seawater and related gases in the DB East District

The different migration times of evaporated seawater in the DB East District could indicate different episodes of natural gas formation. All the gases were enriched in methane and thus exhibited high dryness coefficients (>97%, Table S4). However, the dryness coefficients differed among the wells. As shown in Fig. 9b, t_S and the dryness coefficient are negatively related. These findings indicate that the maturity of the gases that reached different wells varied. The first-charged gas was the least mature and thus had the lowest dryness coefficient (97.1% of DB102, 97.1% of DB101-5, and 97.3% of DB101-1), whereas the second-charged gases were derived from more mature source rocks and thus had relatively high dryness coefficients (97.7% of DB209 and 97.8% of DB202). The third-charged gases were characterized by the highest dryness coefficients (98.4% for DB304 and 98.7% for DB304-1). The distribution of δ¹³C_CH4, which was reported in a previous study, also supported the different episodes of hydrocarbon gases. It is widely accepted that the late-charged gas with the highest maturity has the heaviest δ¹³C_CH4. Here, the δ¹³C_CH4 in the southeast of the DB East District had the heaviest value (-29.7‰, n=3), while it was lightest in the northeast (-30.6‰, n=5)⁷³. The gases in the middle location have an intermediate value of δ¹³C_CH4 (-30.4‰, n=4)⁷³. Furthermore, the migration times of both the gases and the evaporated seawater were strongly related to the local activities of the thrust faults. In the DB East District, three thrust faults developed from northeast to southwest (Keshen No. 1 Fault, Keshen Fault, and Keshen No. 4 Fault; Fig. 1a and Fig. 10). These faults are considered to have formed via compression by the Tianshan Orogenic Belt since the Miocene and have experienced different developmental stages (e.g., sealing or reopening)^34,73. It appears that the opening times of these faults followed the order of the Keshen No. 1 Fault, Keshen Fault, and Keshen No. 4 Fault. Therefore, the natural gas and associated evaporated seawater from the deep Triassic–Jurassic source rocks migrated along the trough faults into the reservoirs at different times.

Fig. 10.

Comigration model of evaporated seawater and gases in the DB East District. The base map was modified from Zhang et al.⁷². The percentages shown in this figure are the dryness coefficients of the natural gases.

The calculated t_S values were not accurate. The t_S values were clearly greater than the charge times of the gases estimated from the inclusion analysis (3–2 Ma)^43,44 and the modeled gas generation times for the source rocks (<5 Ma)^34,36. Furthermore, some t_S values occurred earlier than the formation time of the thrust fault system (Miocene, <23.3 Ma), which was associated with the main upward migration pathway of both hydrocarbons and evaporated seawater. The anomalously higher t_S values could be attributed to the low C₁₂₉ value, which was due to the presence of residual water before the charging of the gas-related evaporated seawater and/or the addition of condensed water. Three kinds of residual water, including connate water when the reservoir sandstones were deposited, meteoric water during the denudation period in the Late Cretaceous and residual brine associated with early charged oils, were presumably displaced by later fluids. Under geological conditions, all three types of fluids were retained in the formation water to some degree and exhibited low C₁₂₉ values as a result of the long-term delay in ¹²⁹I in the reservoirs. In addition, the dilution of condensed water during gas recovery reduced the C₁₂₉ content. Therefore, both the presence of residual water and the dilution of condensed water decreased the C₁₂₉ content and thus increased the calculated t_S value in the dating model (Fig. S2). Nevertheless, the different gas accumulation periods among the wells in the DB East District were revealed by the t_S values of the evaporated seawater.

Geological implications

With petroleum exploration occurring in deep and ultradeep strata, many gas accumulations have been discovered worldwide²⁰. However, unlike crude oils, natural gas is compositionally simple. It is typically composed of several kinds of low-molecular-weight organic (mainly C₁–C₅) and inorganic (e.g., CO₂ and N₂) gaseous compounds. As a product of kerogen maturation and/or oil cracking under high thermal stress, these gaseous compounds provide only limited geochemical information, such as the kerogen type, maturity stage, and gas–gas relationship⁸⁹. The accumulation processes of natural gases are difficult to elucidate clearly from gas geochemistry. Given these factors, attention should be focused on the geochemistry of formation water.

The chemical and isotopic compositions of formation water can not only reveal the origins and evolution processes of the water but also constrain the sources and accumulation processes of the gas. This case study was a preliminary attempt at obtaining such information. However, several problems urgently need to be solved. First, the evolution process of pore water during the maturation of source rocks requires further investigation. Information such as the amount and occurrence state of pore water, relative proportions of pore water and hydrocarbons, and compositions of pore water is key to explaining the coexpulsion behavior of both pore water and hydrocarbons; notably, this information can be used to develop a precise ¹²⁹I dating technique and new tracing indicators. Second, the dating technique should be further improved. Some of the parameters used in the dating equations (e.g., E and P) were assigned on the basis of the user’s experience. The other parameters vary over large scales as a result of fluid mixing and other geological factors. More precise dating results could be obtained if these parameters are further constrained and several different dating methods (e.g., ¹²⁹I and ³⁶Cl) are used together. Finally, the influence of condensed water on produced water needs to be quantitatively assessed to obtain the true chemical and isotopic compositions of the formation water and fully utilize the water data. In summary, by combining the geochemistry of both natural gas and produced water, the overall and precise accumulation processes of natural gas, especially the differentiated accumulation processes in a given region, can be revealed.

Conclusions

In this study, we systemically analyzed the chemical and isotopic compositions of the water produced in a deep BD gas field in northwestern China. The main conclusions are as follows.

1. The formation water mainly consists of halite-dissolution meteoric water and low proportions of iodine-rich evaporated seawater. The former infiltrated the E_1-2km halite during the Kangcun–Kuqa stage (Middle Miocene to Pliocene, 16.3–1.64 Ma). The latter migrated together with natural gas from some layers in the Triassic–Jurassic source rocks that were influenced by seawater invasion. In addition, connate water or other water (e.g., meteoric water during the denudation period of the Late Cretaceous) could have been retained to some degree.

2. Compared with the formation water, the condensed water had lower TDS contents, lower δD and δ¹⁸O values, and different ionic compositions. Most of the produced water displayed characteristics of mixing between formation water and condensed water to different degrees.

3. On the basis of a dating model of ¹²⁹I in the formation water, different episodes of evaporated water and associated natural gas were observed among the wells in the DB East District, although the precise timing for the migration of evaporated seawater was difficult to determine.

Together, the results of this case study add to our understanding of the movement of hydrocarbon-related geological fluids in deep basins and suggest that the chemical and isotopic composition of formation water could be used together with that of gas to constrain the sources and accumulation processes of deeply buried gas.

Author contributions

J.C. wrote the main manuscript text. Y.F., N.C. and J.X. conducted the experimental work. J.C. and Y.F. analyzed the results. H.Z. and T.M. provided the geological information of Bozi-Dabei gas field. W.J. and Y.W. improved the paper’s structure and language. H.Z., T.M., X.H. and P.P. revised the manuscript. All authors reviewed the manuscript.

Funding

This study was financially supported by the National Natural Science Foundation of China (grant nos. 42125304 and 42573026). Theory of hydrocarbon enrichment under multisphere interactions of the Earth (THEMSIE04010104), and the State Key Laboratory of Organic Geochemistry (SKLOG2024-03). This is contribution No.is-3761 from GIGCAS.

Data availability

All data generated or analyzed during this study are included in this published article and its Supplementary Information files.

Declarations

Competing interests

The authors declare no competing interests.