1 INTRODUCTION

The Qaidam Basin (QB) in NW China shows high evaporation rates and very few rainfalls. This hyper-arid environment allows critical accumulations of evaporite salts and the formation of brines due to the interaction of groundwater with the evaporitic deposits and surface evaporation processes (Warren, 2016). The Qarhan Salt Lake (QSL) (Fig. 1) located in the eastern QB is special among other terrestrial basins. For example, it is the largest Quaternary potash deposits in the world with a reverse of 5.40×10⁸ t (Zhang, 1987; Yuan et al., 1995; Cao and Wu, 2004). The sedimentary age of QSL is young (~50 ka) (Zhang et al., 1993; Fan et al., 2014), and its brines and salt formation are buried shallow (20–60 m)(Zhang, 1987; Yuan et al., 1995). The mineral precipitating sequences of the evaporites are relatively simple, which is similar to those of marine MgSO₄-deficient potash deposits in evaporite basins of the world (Lowenstein et al., 1989; Zhang et al., 1993). The recharge sources are diverse, including river water, thermal spring, and Ca-Cl spring (Lowenstein et al., 1989; Zhu et al., 1989; Zhang et al., 1993; Yu et al., 2013). In addition, the Beletan section of the QSL is also an enrichment region of lithium (Li) and boron (B) resources (Fig. 2), which accounts for ~80% of China's total brine lithium resources, together with Yiliping, Xitai, Dongtai, and other modern Quaternary salt lakes (Yu et al., 2013).

On the basis of these particularities, significant research efforts have been done on the origin of the extraordinary potassium (K), Li, and B enrichment and evolutionary history of the QSL, and different hypotheses have been proposed over the years (Chen and Bowler, 1986; Zhang, 1987; Duan and Yuan, 1988; Yuan et al., 1995; Fan et al., 2014, 2018), such as the inheritance and evolution of solutes of Qaidam "mega-paleolakes" in the process of migration from west to east in the QB (Chen and Bowler, 1986), evaporation pumping of intercrystalline brines (Xu et al., 1991), capture of paleolakes in the Kunlun Mountains by rivers and the migration to the Qarhan region in the process of neotectonic movement (~30 ka) (Zhu et al., 1994), weathering input of granite from the Kunlun Mountains (Yang et al., 1993; Yuan et al., 1995; Zhang et al., 2019; An et al., 2021), and the mixing of river waters and Ca-Cl springs (Lowenstein et al., 1989; Zhang et al., 1993; Fan et al., 2018).

However, these views are controversial, such as the view of "mega-paleolake" ignores the fact that B is enriched only in the Bieletan section and significantly less in other sections. Moreover, the time of neotectonic movement (~30 ka) does not correspond to the time of formation of the QSL (~50 ka) (Fan et al., 2014). Therefore, Lowenstein et al. (1989) believed that the mixed view shall be more consistent with the existence of two hydrochemical types (Cl-SO₄ and Ca-Cl) and simple mineral sequences in the QSL (Lowenstein et al., 1989; Zhang et al., 1993; Fan et al., 2018). The mixed-origin view is supported by the simulation results of mixing river and spring water using the Pitzer model (Liu et al., 2002). However, more research is still needed to better understand the formation and evolution of the QSL.

On the other hand, the QSL is the largest potash salt production base of China, and has long focused on the exploration and development of potassium resources. However, its lithium resources attracted more attention in the 21^st century. Due to the lack of intensive study and recognition on the formation and distribution of lithium deposits, the effective protection of co-occurring lithium brines was neglected in large-scale mining of K-containing brines, resulting in serious loss of lithium (Yu et al., 2018). In addition, lithium and boron in the QSL show obvious differential distribution, and the research focused on mainly the Bieletan section, and rarely covered other sections. High strontium (Sr) concentrations of brines in the QSL have also been observed (Zhang, 1987; Fan et al., 2018). Even so, the fact that the brines with high strontium concentrations are neglected in the reality. Therefore, in order to comprehensively understand the distribution, source and rational utilization of resources, the study on distribution and the recharge processes of Li, B, and Sr in the whole QSL are greatly demanded.

Since the source of element supply is closely related to its spatial distribution and enrichment mechanism. In this work, Li-B concentrations of 66 intercrystalline brines, H-O isotopes of 31 brines, and Sr-B isotopes of different recharge fluids (river and spring) in the QSL were determined, and the available published data of potassium and strontium elemental concentrations (Zhang, 1987; Spencer et al., 1990; Zhang et al., 1993; Fan et al., 2018) and multiple isotopes (H, O, Sr, B) in the study area (Qi et al., 1993; Zhang et al., 1993; Xiao et al., 1999, 2018; Tan et al., 2011; Zhang et al., 2013; Cui et al., 2015; Fan et al., 2018; Li et al., 2019, 2021a; Xu, 2021) were collected. We tried to delineate the sources, potential mixing processes, and mineralization in the study area by combing multi-isotopic and elemental composition features from the perspective of recharge sources, and compare them with geological analogues in the evaporate basins. This work is beneficial to utilization and extraction of multiple resources and protection of different resource grade in the QSL.

2 GEOLOGICAL AND HYDROLOGICAL SETTING

The rhombic-shaped QB, located at the north-eastern Qinghai-Tibet Plateau, is a closed, fault-depressed intermountain basin with an area of 120 000 km². The basin tectonically belongs to the Tethys tectonic domain in the Central Asia arid region. The Altyn fault, Kunlun fault, and Northern Qaidam fault constrained the QB (Molnar and Tapponnier, 1975; Yin, 2000; Yin and Harrison, 2000). Geological data have shown that a large number of the Cenozoic sediments developed in the QB, and the thickness in some places reached ~14 km (Yin et al., 2007, 2008; Meng and Fang, 2008; Cheng et al., 2016, 2018). Because of collision orogeny, the Altyn Tagh, Qilian Mountains, and the Eastern Kunlun orogenic belts on the periphery of QB recorded a large amount of Early Paleozoic-Early Mesozoic tectono-magmatic events (Cheng et al., 2017). The Qaidam basements that are covered by as thick as ~14-km Cenozoic sediments are composed of the Early Neoproterozoic, Early Paleozoic, and Late Paleozoic to Mesozoic granitoids (Cheng et al., 2017). A number of uplifts, depressions, and sub-faults are formed in the QB due to the heterogeneity of the long-term repeated activities of deep and large faults, dividing it into three structural units: the eastern and western depressions, and the northern margin fault-fold belt (Zhang, 1987).

The QSL located at the eastern depression, is currently the lowest depocenter and the largest sub-basin in the QB (Fig. 1), measuring 168 km east to west and 20–40 km north to south and covering an area of 5 856 km² (Yuan et al., 1995; Huang and Han, 2007). The south of QSL is the Kunlun Mountains, and the east is the Ela Mountains, which are separated from the Chaka-Gonghe basins. The Lüliang, Xitie, and Amunik Mountains are to the north of QSL (Fig. 1) (Yuan et al., 1995). The west of QSL is the Sebei anticline that separates it with the Dongtai, Xitai, and Yiliping salt lakes. The QSL can be divided into four sections from west to east: Bieletan (BLT), Dabuxun (DBX), Qarhan (QRH), and Huobuxun (HBX) (Figs. 2 & 3a).

According to the hydrologic systems, QSL contains 10 shallow perennial and ephemeral saline lakes, which can be classified into two main brine types: Cl-SO₄ and Ca-Cl types (Fig. 2) (Zhang, 1987; Yang et al., 1993; Yuan et al., 1995; Fan et al., 2018). The Cl-SO₄-type brines can be found in the southern and southwestern QSL, including Senie, Xiaobiele, Dabiele, Dabuxun, Tuanjie, and South Huobuxun lakes. The Ca-Cl type brines are mainly located in the northern and northeastern QSL, including Dongling, Xiezuo, and North Huobuxun lakes (Figs. 2–3) (Zhang, 1987). These two types of intercrystalline brines, Cl-SO4 and Ca-Cl types, can be found at QSL. The spatial distribution is consistent with that of surface brines or lakes (Fan et al., 2018). The catchment area of QSL is 132 000 km², and nearly 20 rivers flow into the playa (Fig. 1). The rivers mainly originate from natural precipitation and meltwater from the Kunlun Mountains in the south QSL, whilst the Amunik and Xitie Mountains in the north QSL are below the snow level and has only one seasonal Quanji River (Yuan et al., 1995). The Golmud River, which originates from Kunlun Mountains (Fig. 1), has the largest flow of 7.98×10⁸ m³/a and discharges into the Dabuxun Lake (Zhang, 1987). The Senie Lake in Bieletan (BLT) section is fed by the Wutumeiren and Zaohuo rivers. The Dabiele, South Huobuxun, and North Huobuxun lakes are fed by the Qingshui, Nuomuhong, and Qaidam Rivers, respectively (Fig. 1). Ca-Cl springs from a karst zone near the northern margin are another main inflow replenishing the QSL (Lowenstein et al., 1989; Zhang et al., 1993; Lowenstein and Risacher, 2009; Fan et al., 2018) (Fig. 2). At the karst zone, Ca-Cl springs discharge along a deep-seated northwest-oriented fault zone mapped over a distance of several hundred kilometres (Vengosh et al., 1995). The Ca-Cl springs are rich in Ca²⁺, Na^⁺, and Cl^-, with minimal SO₄^2- and almost negligible HCO₃^- and CO₃^2-(Lowenstein et al., 1989; Spencer et al., 1990; Zhang et al., 1993).

In addition, the upper reaches of Hongshui River near Mount Bukadaban in the eastern Kunlun Mountains are tectonically active, with fracture formation, and hundreds of Li- and B-rich thermal springs (Hu, 1997; Pang, 2009) from this area flow into the Hongshui River, and then into the Nalinggele River and supplies the terminal salt lakes (Yiliping, Xitai, Dongtai, and BLT section of QSL) (Figs. 1–2) (Yu et al., 2013; Li et al., 2021b). By contrast, to the north QSL, B- and Li-rich thermal springs emerge from a major WNW-ESE-trending fault system between the Qaidam granite and Paleozoic sediments and is surrounded by Mesoproterozoic metamorphic rocks in the Qilian Mountains. The thermal springs provide vital inflow water and boron, lithium resources to the lake water in this area, such as Da Qaidam Lake (Kong et al., 2021; Liu et al., 2022). These thermal springs form spatially a north-south echo in the QB.

3 SAMPLE AND ANALYTICAL METHOD 3.1 Sample collection

Sixty-six intercrystalline brine samples from BLT, DBX, QRH, and HBX sections of the QSL were collected during 2012–2014 (Fig. 3a). Three river waters of Nalinggele River and four Ca-Cl springs from the northern QSL have also collected. All brine samples were collected in acid-washed 500-mL low-density polyethylene (LDPE) bottles that were washed by dilute hydrochloric acid and rinsed with environmental water. The samples were filtered with a 0.45-µm polypropylene membrane into an acid-washed 125-mL LDPE bottle within 72 h of collection. The bottles were then sealed and sent to the laboratory for isotope and element analyses.

3.2 Analytical method

The lithium and boron resource elements of the intercrystalline brines in the QSL were analysed in this study at the Qinghai Institute of Salt Lakes, Chinese Academy of Sciences. All the analyses were completed within a week after sampling. The lithium and boron concentrations of the samples were determined with an inductively coupled plasma optical emission spectrometer (ICP-OES, ICAP6500 DUO, Thermo Scientific, USA). The analytical precision for these elements is more than ±5%.

The thirty-one brine samples were prepared for analysis for the hydrogen (H) and oxygen (O) isotopes at Open Laboratory for Isotope Geochemistry, China Geological Survey. The H-O isotopes were determined via H₂ and CO₂ on a MAT253 Isotope Ratio Mass Spectrometer (IRMS, Thermo Fisher Scientific Inc., USA) and by zinc reduction and CO₂-H₂O equilibrium. The Vienna Standard Mean Ocean Water was taken as the standard materials for the H-O isotopic analysis with the analytical precisions of 1‰ and 0.1‰. Three river-water and four Ca-Cl-spring samples were prepared for analysis of strontium and boron isotopes, respectively. Strontium isotope analyses were performed on a MAT261 Thermal Ionization Mass Spectrometer in the Open Laboratory for Isotope Geochemistry, China Geological Survey, at analytical precision of ±1.0×10^-6 (2σ, SE). Measurement of δ¹¹B was made with the thermal ionization mass spectrometer (Triton) at Qinghai Institute of Salt Lakes, Chinese Academy of Sciences. The B Isotope Reference Material was NIST SRM951 H₃BO₃, the test accuracy was better than 0.2‰. Sample handling and testing methods refer to Xiao et al. (1988) and Wang et al. (2002).

4 RESULT 4.1 K-Li-B-Sr elemental abundance of the brines in the QSL

The sample numbers, regions, and potassium, lithium, boron, and strontium concentrations in the intercrystalline brines in the QSL are shown in Table 1. Although potassium and strontium concentrations of brines has been reported (Fan et al., 2018), the sedimentary pattern of potassium and strontium resources has been discussed in this study. Significant differences in potassium concentration can be observed in the different sections of the QSL. The average values of BLT, DBX, QRH, and HBX are 16.32, 9.87, 4.34, and 4.14 g/L, respectively. However, DBX section has the highest potassium concentration, followed by BLT section, especially the northern portion of these two sections (Fig. 4a). The average strontium concentration is highest in QRH section (251.65 mg/L), slightly low in HBX section (191.38 mg/L) and significantly low in DBX (45.39 mg/L) and BLT (29.08 mg/L) sections (Fig. 4b; Table 1). The intercrystalline brines in BLT section have the highest lithium and boron concentrations of 183.25 and 1 704.74 mg/L, respectively (Fig. 4c–d). The lithium and boron concentrations diminished eastward.

In addition, previously reported potassium, lithium, boron, and strontium ionic concentrations of the Ca-Cl springs, river waters, and thermal springs in the QSL (Zhang, 1987; Yang et al., 1993; Zhang et al., 1993; Tan et al., 2012; Fan et al., 2018; Li et al., 2021a) are illustrated in Supplementary Table S1. The Ca-Cl springs have relatively low potassium concentration from 0.16 to 12.08 g/L, averaging 4.00 g/L, which is lower than those of the intercrystalline brines. The strontium concentrations of the Ca-Cl springs are relatively high, ranging from 341.28 to 663.28 mg/L on average of 512.61 mg/L. In the river waters, the potassium and strontium concentrations are relatively low, ranging 2.35–88.37 mg/L and 0.061–1.02 mg/L, respectively (Supplementary Table S1).

4.2 H-O-B-Sr isotopic compositions of the brines in the QSL

The δD and δ¹⁸O of intercrystalline brines in the QSL ranges from -56.6‰ to -8‰ for δD and from -5.46‰ to 3.36‰ for δ18O (Fig. 5; Table 1). The isotopic compositions in the QSL are heterogeneous among different sections. In BLT and DBX sections, the brines have relatively high δD and δ¹⁸O, with average δD of -23.8‰ and -28.5‰, and δ¹⁸O of 0.65‰ and 1.62‰, respectively (Fig. 5; Table 1). Meanwhile, the brines from QRH section show lower δD and δ¹⁸O values, averaging -38.4‰ and averaging -0.12‰, respectively. The δD and δ¹⁸O values of the brines from HBX section vary between -56.6‰ and -17.6‰, -5.46‰ and 2.08‰, respectively (Fig. 5; Table 1). The previously reported δD and δ¹⁸O values of the river and Ca-Cl springs in the QSL are also collected and shown in Supplementary Table S2. The H-O isotopes of the river waters of catchment area are low, with averages of -64.13‰ and -9.03‰, and those of Ca-Cl springs are relatively higher, with an average of -49.47‰ and -1.56‰ (Fig. 5; Supplementary Table S2). This high oxygen isotopic values are caused by the water-rock reaction between the groundwater and the surrounding rock in the aquifer (Zhang et al., 1993; Yi, 2017).

The δ¹¹B in the river waters feeding the QSL is relatively low (-3‰–1.5‰) (Qi et al., 1993; Xiao et al., 1999) and plotted in low B isotope end-member. On the contrary, the δ¹¹B values are relatively high in Ca-Cl springs, ranging from 13.8‰ to 28.9‰ and located in the high B isotope end-member. The δ¹¹B values in the terminal salt lakes are between the two end-members (Fig. 6a; Table 2). The distribution pattern is similar to that of ⁸⁷Sr/⁸⁶Sr in the study area (Fig. 6b). The ⁸⁷Sr/⁸⁶Sr value of the terminal salt lakes are between the river water with low ⁸⁷Sr/⁸⁶Sr end-member (0.710 8–0.711 60) (Fan et al., 2018) and the Ca-Cl springs with high ⁸⁷Sr/⁸⁶Sr end-member (0.711 58–0.711 80) (Qi et al., 1992; Fan et al., 2018).

5 DISCUSSION

The geological survey has indicated that two water sources, rivers from the southern Kunlun Mountains and Ca-Cl springs from the northern fault zone, recharged solutes of lake water and altered the water chemistry of the brine deposits in the QSL (Lowenstein et al., 1989; Zhang et al., 1993; Yuan et al., 1995; Fan et al., 2018). The CaSO₄-HCO₃ diagram of surface brines and river waters presented firstly two water end-members (river and Ca-Cl spring) and spatial mixing process happened in the QSL (Lowenstein et al., 1989). The ratios of strontium concentrations to ⁸⁷Sr/⁸⁶Sr value (Fan et al., 2018) and boron concentrations to δ¹¹B value (Qi et al., 1993; Xiao et al., 1999; Ma et al., 2015) in different waters (river water, intercrystalline brine, Ca-Cl spring, thermal spring) further depicted the recharge sections and recharge elements of different waters. On the one hand, the H-O-Sr-B isotopic values of rivers, brines, Ca-Cl springs, and thermal springs emphasize the mixing characteristics of brines in the different sections and the peculiarity of Ca-Cl springs that is deep source and strong water-reaction (Fig. 5). On the other hand, three clusters of high potassium and low calcium, high strontium and high calcium, and high lithium-boron and low calcium elements can be seen in Fig. 7. These results indicate that different resource elements are controlled by different recharge water sources. We found that high potassium values occurred in BLT and DBX sections (Figs. 4a & 7a) and far away from the high Sr-Ca clusters (Figs. 4b & 7b). High Li-B values happen in BLT section, but do not include DBX section (Fig. 4c–d). These comparisons show that the potassium, strontium, lithium, and boron resource elements in the QSL are controlled by different recharging sources or recharging processes.

5.1 Recharge and mixing of the river waters and Ca-Cl springs limit K-Sr in the QSL

In the QSL, the potassium concentrations of the intercrystalline brines are significantly enriched in the northern part of BLT and DBX sections (the peak value is in DBX section), and the average value gradually decreased eastward from BLT to HBX (Fig. 4; Table 1). This potassium elemental pattern in brines is also in accordance with the distribution of the K-bearing evaporites (especially carnallite salts in DBX section) in the QSL (Zhang, 1987; Yuan et al., 1995). The recharge source of high potassium of brines in BLT and DBX is obvious different from that of high Sr-Ca concentrations of brines in QRH and HBX (Fig. 7a–b), indicating that Ca-Cl springs are not main source for potassium elements in the study area. Yuan et al. (1995) pointed out that river waters were the main solute source (including K) in the QSL. The high-flux potassium in rivers is attributed to the input of weathering and denudation of high potassium granite outcrops (3.5%) (Zhang, 1987; Yang et al., 1993; Yuan et al., 1995; Zhang et al., 2019). Zhang et al. (2019) additionally calculated that the total amount of KCl brought by rivers from Kunlun Mountains was 12.73×10³ t/a (Fig. 8). But the weathering input of rivers cannot explain the highest potassium value and carnallite salts occurred in the north part of Dabuxun Lake in DBX section (Fig. 2) because of more K-flux inputs (7 160 t/a; Fig. 8) in BLT section. The mixing model of river water and Ca-Cl springs seems to provide a reference for interpreting differential formation of potassium resources in the QSL (Lowenstein et al., 1989). On the one hand, this mixing model proposed that excessive Ca recharged by the Ca-Cl springs consumed SO₄^2- in lake waters and altered the water chemistry of brines, and correspondingly formed MgSO₄-deficient evaporites (calcite, gypsum, halite, sylvite or carnallite and bischofite) during the evaporation process. The potash minerals are more easily precipitated during the simple precipitating mineral sequence in the evaporite strata. On the other hand, the topography of the Bieda uplift (Fig. 8) blocked the hydrological connectivity between BLT section and DBX-QRHHBX sections, resulting in a relatively closed BLT sub-basin. Compared with those in BLT, therefore, a good deposition space (thicker evaporite strata, ~55 m) and hydrological connectivity in DBX (Fig. 8) favored the evaporite deposition in relatively deep sub-basin and enough mixing of Ca-Cl springs. Therefore, the recharge source and mixing of Ca-Cl springs limit potassium resource distribution in the QSL.

Compared with potassium element, high strontium in brines is closely related with Ca-Cl springs in the QSL (Fig. 7b) and its distribution pattern is relatively simple. The strontium concentrations of the Ca-Cl spring overflowing from the northern fault zone are higher, and the closer to the fault zone, the higher the strontium concentrations are, and vice versa. In addition, a favorable positive correlation between calcium and strontium concentrations indicates that these two elements are from the same source (Fig. 7b). The lowest δD values and high ⁸⁷Sr/⁸⁶Sr in QRH section (Fig. 5, the closest to the spring waters; Fig. 6b) also demonstrate that recharge of Ca-Cl springs controls strontium element distribution of brines in the QSL.

5.2 Recharge of thermal spring limits Li-B in the QSL

The lithium and boron concentrations in brines are the highest in BLT and gradually decreased eastward sections in the QSL (Fig. 4). High lithiumboron and low calcium clusters in BLT section (Fig. 7c–d) indicate a weak mixing recharge of CaCl springs. The δD and δ¹⁸O values of the intercrystalline brines in BLT section are located closest to the local evaporation line (LEL) (Fig. 5) and the low δ¹¹B value of BLT is similar to that of HN rivers (the Hongshui River is a tributary of the upper reach of Nalinggele River that has developed a mixing river, termed Nalinggele-Hongshui (N-H) rivers here) waters (Fig. 6a), indicating that its water sources and boron elements are controlled by river waters. The Nalinggele River and Wutumeiren River are the main water sources to BLT section in the QSL (Fig. 2). A hydrothermal field evolved in the Hongshui River valley due to the neotectonic movement of the Kunlun fault and frequent earthquakes (Tapponnier et al., 2001). The hot water at high temperatures exceeding 300–350 ℃ leached the volcanic rock, and abundant lithium and boron are carried and poured out onto the surface through thermal springs, which is presumably the main origin of lithium and boron in the Hongshui River and terminal salt lakes (including the BLT, Dongtai, Xitai, and Yiliping salt lakes) (Yu et al., 2013). The lithium and boron concentrations of thermal springs can reach 96.0 and 180.0 mg/L, respectively (Hu, 1997). The average lithium concentration of the HN-rivers water near out of Kunlun Mountains reached 0.727 mg/L, which is about 20 times higher than that of Golmud River waters (Yu et al., 2013, 2018). The salinity and Li-B concentrations gradually decrease from upstream to downstream in the H-N rivers (Zhu et al., 1990), which suggests that the weathering and erosion of surrounding rocks and Quaternary sediments contributed less to the input of lithium-boron elements. The similar high lithium concentrations of H-N rivers and thermal springs (Fig. 7c) further constrained the high lithium concentration in brines in BLT and from the recharge of thermal springs in the Kunlun Mountains. Therefore, the flowing water of H-N rivers (i.e., from thermal springs) provide sufficient lithium and boron for river waters and terminal salt lakes.

5.3 The formation and evolution of two springs constraint fundamentally on different resource elements of terminal salt lakes

On the basis of the discussion above mentioned, we concluded that different recharge processes limit the sources and distribution of potassium, lithium, boron, and strontium elements, making these resources coexist in the QSL. Due to the differences in the spatial distribution of recharge sources and regional environment, the strontium controlled by recharge of Ca-Cl springs is mainly concentrated in the QRH section, and the potassium source is from weathering and leaching of K-rich granite in the Kunlun Mountains, but its enrichment and the precipitation of potash salts are controlled by the mixing of river water and Ca-Cl springs, especially the latter. Therefore, the enrichment of potassium and strontium resource elements in the QSL are related closely to the recharge and mixing of Ca-Cl springs. By contrast, lithium and boron controlled by the recharge of thermal springs are mainly enriched in BLT section, the mineralization of lithium and boron resources is mainly related to the evaporation and concentration. Different spring sources carry different resource elements into the terminal salt lakes. Hence, the source and distribution of multiple resources of QSL are actually constrained by two different springs, that is thermal springs and Ca-Cl springs. The fundamental reason for difference bearing resource elements between these two springs lie in its formation or evolution.

These two springs were formed on the basis of the evolution of proto-Qaidam Basin (Fig. 9a). The Qaidam block is a tectono-magmatic rejuvenated craton. The surrounding of the block developed multiple large fault and small basement fault (Fig. 9a). Along with the collision of Indo-Asian plate since the Cenozoic (Cheng et al., 2019, 2021), the periodic uplift of Altyn orogenic belt, Eastern Kunlun orogenic belt and Qilian Mountain orogenic belt has aggravated the development of the QB and formed several source-sink systems between the mountains and the basin. The eastern Kunlun and terminal K-Li-rich salt lakes in the eastern-central QB are an important and typical representative (Fig. 9b). The thermal springs produced by waterrock reaction along Pliocene-Quaternary volcanic centers in the Kunlun Mountains (Tapponnier et al., 2001). Therefore, the formation of this thermal springs may be related to high-temperature watervolcanic rock reaction caused by deep heat sources in the area. And a large amount of lithium and boron are leached and transported through active hydrothermal channels, and spouted to the surface by thermal spring vents, which does not exclude the involvement of deep magmatic fluids (Yu et al., 2013) (Fig. 9b).

On the contrary, many anticlines and faults formed droved by north-south stress along with Indo-Asian plate collision. The Ca-Cl springs were generated through a displacement fault (Lowenstein et al., 1989) due to an unconformity between Paleogene-Neogene strata and Quaternary strata (Fig. 9b). Therefore, the formation of Ca-Cl springs in the northern QSL is associated with the anticline of Paleogene and Neogene sediments, and similar to that of oilfield waters in the western QB. The high δ¹⁸O values of brines in the QSL (Fig. 5) are robust evidence lines. It is formed by precipitation infiltration along deep faults and water-rock interaction with Paleogene and Neogene sediments. The high Ca in Ca-Cl springs and its evolution process are related to the metasomatism of calcite by dolomite and the dissolution of calcite (Yuan et al., 1995; Fan et al., 2018) (Fig. 9b), In addition, some evaporites in the Cenozoic sediments, e.g., Ca-rich evaporites (gypsum, anhydrite) and Cl-rich evaporites, e.g., halite (Guo et al., 2017) can also provide some Ca-Cl.

5.4 Analogues of recharge model constraint the resource elements in the evaporite basins

Many of the evaporative salt basins are known for their enrichment of beneficial elements. There is consistency between them, that is, the formation of resources is controlled by their replenishment patterns. K-rich salt lakes, such as Lop Nur Salt Lake (Ma et al., 2022) and Kunteyi Salt Lake in China, and Salar de Atacama in Chile, South America (Lowenstein and Risacher, 2009), etc., are all found to have the participation of Ca-Cl-type waters (Fig. 10a–b). The mixing of river waters and Ca-Cl springs is the key to potassium enrichment. As the second largest production base of potash salt in China, Lop Nur Salt Lake, the concentrations of potassium in brines increases exponentially (from 4 to 8 g/L) (Zhang et al., 2021a, b) when the hydrochemical type transforms from sulfate type into Ca-Cl type (indicating the gradual mixing process of deep Ca-Cl type water and upper stratum water).

Li-B-rich salt lakes, including the Da Qaidam Lake, Xiao Qaidam Lake, and Zhabuye Salt Lake of the Qinghai-Tibet Plateau, Salar de Atacama and Guayatayoc playa lake of South America, etc. (Liu et al., 2007; Zheng and Liu, 2009; Godfrey et al., 2013; Steinmetz, 2017; Kong et al., 2021), are commonly recharged by thermal springs. An existing good analogue of source-sink recharge process of thermal springs can be observed in the Da Qaidam Lake to the north of QSL (Fig. 9b). This recharge of Li-B resource to the Da Qaidam Lake and the QSL forms spatially a north-south echo in the QB. In addition, the scene of source-sink recharge processes in the QB are coincident with those in Xizang and the South American "Lithium Triangle" (Fig. 10c–d). This thermal spring recharge, corresponding to the enrichment of elements, from thermal springs to river, and then to terminal salt lakes can be concluded in Fig. 10c–d. For example, thermal springs (currently active in the Coranzuli crater) with lithium concentrations of ~17 mg/L supply lithium to the Las Burras River, resulting into the Las Burras River provides the most concentrated lithium input (3.75 mg/L) and boron input (20 mg/L) for the Guayatayoc playa lake system (Steinmetz, 2017). At present, the brines in Guayatayoc playa contains up to 125 mg/L of lithium.

Therefore, two recharge models are responsible for multiple resource-elements in terminal salt lakes in the world. The mixing pattern of river and Ca-Cl type water promoted the significant enrichment of potassium resource in brines. The lithium and boron resources in salt lakes are controlled by recharge of thermal springs and enriched by evaporation and concentration. The study on the sources and metallogenic models of potassium, strontium, lithium, and boron in the QSL has certain enlightenment significance for the mineralization of salt in other evaporative basins in the world.

6 CONCLUSION

Based on K-Sr-Li-B elements and H-O-Sr-B isotopes data from of different waters (river, spring and brines) from the Qarhan Salt Lake from the Qaidam Basin, the following conclusions are drawn:

(1) The concentrations and spatial distribution of the potassium, strontium, lithium, and boron elements in the QSL are diverse. The high potassium and strontium concentrations are mainly distributed in DBX and QRH sections, respectively. Strontium is mainly discharged by upwelling Ca-Cl springs, and potassium resources are from the weathering and leaching of the K-rich granites in the southern Kunlun Mountains and significantly enriched in the brine deposits by the mixing of the rivers and Ca-Cl springs. Whilst lithium and boron resources are mainly enriched in BLT section and primarily originated from thermal springs, where they were enriched by evaporation and mineralization.

(2) Two different recharge springs depict and control the multiple resource elements in the Quaternary salt lakes in the QB. The fundamental reason for difference bearing resource elements between these two springs lies in its formation or evolution.

(3) Some analogues of recharge models limit the resource elements of the QSL and other salt lakes (Da Qaidam, Lop Nur, Zabuye, Atacama, and Guayatayoc) in the world provides a reference for the resource exploration in deep formation waters in the evaporite basins.

7 DATA AVAILABILITY STATEMENT

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

8 ACKNOWLEDGMENT

We thank the reviewers and editor-in-chief for their helpful suggestions and comments. We also thank from Qinghai Institute of Salt Lakes, Chinese Academy of Sciences for their help in field works. We also acknowledge the assistance and support of sample analysis from the Salt Lake Chemical Analysis Center of Qinghai Institute of Salt Lakes, Chinese Academy of Sciences.

Electronic supplementary material

Supplementary material (Supplementary Tables S1–S4) is available in the online version of this article at: https://doi.org/10.1007/s00343-023-2258-7.