Multi-Stable Isotope Constraints on the Sources and Evolution of Potash-Forming Fluids in the Mahai Basin, Qinghai–Tibetan Plateau

Abstract

The Mahai Basin (MHB), situated in the northern Qaidam Basin on the Qinghai–Tibetan Plateau, hosts significant Quaternary potash resources. Nevertheless, the sources and evolutionary pathways of potash-forming fluids remain controversial. In this study, a comprehensive multi-isotope dataset and online-first publications spanning the period from 2015 to 2025 were compiled to constrain the end-member characteristics and evolution of brines in the MHB. δD-δ¹⁸O indicates that the initial fluids were derived mainly from Qilian Mountains precipitation and snowmelt, delivered via surface runoff and concentrated through prolonged evaporation under arid, semi-closed conditions, forming a river-lake-brine evolution sequence. δ⁷Li (+7‰ to +40‰) systematically increases with salinity and K content, reflecting long-term low-temperature water–rock interactions and selective ⁶Li adsorption by secondary clays, while deep Ca-Cl brines represent highly evolved endmembers. Elevated ⁸⁷Sr/⁸⁶Sr ratios (0.7113–0.7122) confirm silicate weathering contributions, with intercrystalline brines acting as key intermediate end members. B, S, and Cl isotopes further highlight deep fluid ascent along faults and anticlines, driving K co-enrichment, while sandy–gravel brines inherit highly evolved paleo-lake signatures. These multi-isotope constraints define an integrated evolutionary model involving surface runoff recharge, evaporation-driven concentration with water–rock interaction, deep fluid mixing, lateral migration, and final potash precipitation.

1. Introduction

Potassium (K) is an essential nutrient for plant growth and a key component of agricultural fertilizers, playing a critical role in global food security [1]. Globally, more than 90% of K production is consumed by the fertilizer industry, with the remainder used for industrial purposes [1,2]. Although China ranks fourth worldwide in terms of K resource reserves [3], domestic production satisfies only approximately 50% of national demand, resulting in a strong reliance on imports [4]. Consequently, the identification and development of new K resources are of strategic importance for ensuring sustainable agricultural development and national food security in China [5,6].

K resources in China are mainly distributed in the Qaidam Basin (QB), Lop Nur in Xinjiang, salt lakes of the Tibetan Plateau, and the Mengyejing area in Yunnan [2,7]. Among these regions, the QB, located on the northeastern margin of the Qinghai–Tibet Plateau, is a Mesozoic–Cenozoic intermontane sedimentary basin containing 33 salt lakes. Proven K reserves in this basin account for more than 65% of the national total [3,4,8]. The basin hosts abundant solid and liquid K resources [9] and represents the most important base for K exploitation in China [10]. Therefore, investigations into the K metallogenic mechanisms of the QB are of fundamental significance for advancing theoretical models of continental K mineralization.

In the QB, K is predominantly concentrated in salt lake brines [9,11]. The Mahai Basin (MHB), located in the northern QB, hosts the second-largest medium-scale K deposit after the Qarhan Salt Lake [12]. K mineralization in this area occurs in solid–liquid assemblages developed at surface to shallow subsurface levels [13,14,15]. Multiple types of K-rich brines are present, including confined brines, sandy–gravel brines, anticline-related brines, and oilfield brines [15,16,17]. Different brine types in the MHB exhibit distinct hydrochemical characteristics [7,17,18], making it an ideal natural laboratory for integrated studies of the hydrogeochemical characteristics and formation processes of K-rich brines [5]. These features provide fundamental insights into the deposition and provenance of brine and evaporites. Consequently, investigations into the genesis and occurrence mechanisms of K-rich brines in the MHB are of particular importance, especially for constraining the sources of ore-forming materials and refining metallogenic models for K resources in the Qaidam Basin.

In recent years, extensive research has been conducted on the genesis, occurrence patterns, spatial distribution, and exploration strategies of potash deposits in the MHB. These studies encompass brine hydrochemical classifications [5,10,19], mineralogical characteristics [7,12,20], and simulations of evaporative precipitation processes [21]. As a result, abundant datasets have been accumulated and relatively mature theoretical frameworks have been established [15,19,22]. With the rapid development of non-traditional stable isotope techniques, multi-isotopic systems have become powerful tracers for investigating material sources and enrichment mechanisms in salt lake systems [3,9,19,21,23]. Isotopic studies have significantly advanced our understanding of fluid sources, evolutionary pathways, and metallogenic processes in the MHB [7,8,21,24,25].

Despite these advances, several key issues remain insufficiently synthesized, including the identification of fluid end members, quantitative evaluation of their relative contributions, and the mechanisms governing multi-source fluid mixing. This study aims to systematically review recent stable isotope geochemical studies in the MHB, with a focus on: (1) the isotopic compositional characteristics of different brine types; (2) constraints imposed by isotopic evidence on fluid sources; and (3) isotopic implications for fluid evolution processes. This synthesis is expected to enhance our understanding of ore-forming fluid end members and their evolution in the MHB, thereby providing a robust theoretical basis for elucidating potash metallogenic processes in the MHB and the Qaidam Basin as a whole.

2. Geological and Hydrological Settings

The MHB, situated in the northeastern QB, is an inland sedimentary basin that formed during the Cenozoic era in response to the tectonic uplift of the Tibetan Plateau (Figure 1a,b) [10,26]. It is bounded by the Saishiteng Mountains to the north, the Altun Mountains to the northwest, and the Lenghu and Nanbaxian anticline belts to the southwest and south, respectively. The basin covers an area of approximately 3700 km² and is structurally controlled by rhombic fault systems [13,20]. The MHB is characterized by an extremely arid climate, with a mean annual precipitation of only 29.61 mm and a potential evaporation rate of up to ~3040 mm/yr [5]. Most of the basin surface is covered by dry salt flats, with a total evaporite-bearing area of approximately 843 km² [13]. The southeastern part of the basin represents the lowest topographic area and hosts modern lakes and marshes, whereas the northwestern part forms a narrow valley connecting the Niulang-Zhinu Lakes (NLZNL) [21,27].

Quaternary strata in the MHB are dominated by lacustrine deposits (Figure 1c), including Early Middle Pleistocene, Late Pleistocene, and Holocene sediments, which are generally unconformable with Paleogene–Neogene anticline structures along the basin margins [10]. The marginal hills are composed mainly of the Paleogene Lower Ganchaigou Formation (E₃) and the Neogene Upper Ganchaigou Formation (N₁) and Upper Youshashan Formation (N₂). The E₃ unit consists predominantly of deep lacustrine mudstone, siltstone, and marly limestone; the N₁ unit comprises semi-deep lacustrine mudstone and sandstone; and the N₂ unit is characterized by shallow lacustrine gray mudstone, sandy mudstone, and locally developed gypsum layers [28]. The Saishiteng Mountain range along the northern basin margin is composed mainly of the Precambrian Dakendaban Group and the Lower Paleozoic Tanjianshan Group. Multiple phases of mafic, intermediate, and felsic intrusive rocks are widely developed, dominated by granodiorite and monzogranite, reflecting tectono-magmatic activities from the Caledonian to Yanshanian periods [29]. The Quaternary deposits within the basin are dominated by lacustrine–evaporitic clastic sediments, forming saline successions composed of interbedded silt, clay, and halite. The principal minerals include halite, gypsum, mirabilite, and various K minerals. K brine deposits are widely distributed throughout the basin, whereas solid K deposits are mainly concentrated in the northeastern part of the MHB [5,6,14].

The MHB represents a regional convergence and discharge zone for surface water and groundwater. Brines are mainly hosted in Quaternary salt layers and marginal clastic sediments [10,26]. Influenced by neotectonic activity, the subsidence center, evaporite development, and brine resources are predominantly concentrated in the northeastern part of the basin [30]. Dezong Mahai Lake (DZMHL), Balun Mahai Lake (BLMHL), and NLZNL, together with their surrounding salt flats, constitute an integrated surface brine-intercrystalline brine system [5]. Based on occurrence medium, burial depth, and hydrochemical characteristics, brines in the MHB can be classified into surface brine, intercrystalline brine, confined brine, sand-gravel brine, and anticline-related brine. The K⁺ concentration across these water bodies varies significantly from 0.01 to 20.54 g/L (Table S2). Specifically, the deep confined brines exhibit the highest potassium enrichment, whereas lake waters remain relatively dilute. In terms of hydrochemical facies, surface brine and intercrystalline brine are dominated by the Cl-SO₄ type and are mainly distributed in the central northern basin and the DZMHL area. Sandy–gravel brine and anticline-related brine are characterized by the Ca-Cl type and occur in the piedmont alluvial fans of the Saishiteng Mountains and the Lenghu anticline belt, respectively. In addition, some surface water and shallow groundwater exhibit an Na-HCO₃-SO₄ hydrochemical type [26,31]. Deep confined brines are mainly distributed below 18.5 m within multilayer Quaternary confined aquifer systems. Their spatial distribution is jointly controlled by active faults and unconformities between Paleogene–Neogene strata and Quaternary deposits [21,28]. Regional water sources are primarily derived from snowmelt and groundwater recharged from the Saishiteng Mountains and surrounding highlands, as well as runoff from the Yuka River. The Yuka River is the only perennial river in the region; it directly recharges DZMHL and indirectly influences the water volume and hydrochemical composition of BLMHL through groundwater overflow.

3. Data Collection

A systematic collection of stable isotope datasets was conducted to represent various fluid end-members in the MHB, encompassing peer-reviewed and online-first publications spanning the period from 2015 to 2025. Literature searches were conducted across multiple bibliographic databases, including Web of Science, Google Scholar, and the China National Knowledge Infrastructure (CNKI), using a systematic search logic with Boolean operators (e.g., “AND” and “OR”). Key search terms encompassed geographic identifiers (“Mahai”, “Qaidam Basin”), target substances (“potash”, “brine”), and specific isotopic tracers (“stable isotopes”, “isotope geochemistry”).

Compiled isotope data and associated metadata were extracted directly from published tables, text, and Supplementary Materials. In instances where numerical data were unavailable in the manuscript body and only presented graphically, data were digitized using the GetData Graph Digitizer to ensure high fidelity to the original results. For each entry, we recorded comprehensive information, including sample types, isotope compositions, reference standards, and the specific laboratories or analytical facilities involved. No recalculation or normalization of isotope values was applied beyond the original publications to maintain data integrity.

To ensure comparability among datasets from various sources, data were screened based on reported analytical techniques (e.g., IRMS, MC-ICP-MS, and TIMS) and uncertainty ranges. Values lacking detailed methodological descriptions or analytical precision were excluded. It was maintained that the original authors and journal editors had implemented rigorous quality assurance measures; nonetheless, cross-isotope consistency checks were performed where it was feasible to further validate the data. As this study provides a literature-based synthesis, the analysis emphasizes comparative evaluation—including isotope ranges, clustering patterns, and cross-isotope correlations—rather than formal statistical modeling. The resulting dataset encompasses major isotopic systems essential for tracing fluid evolution in the MHB, including hydrogen–oxygen isotopes (δD-δ¹⁸O, n = 214, Table S1), lithium isotopes (δ⁷Li, n = 70, Table S2), strontium isotopes (⁸⁷Sr/⁸⁶Sr, n = 46, Table S3), boron isotopes (δ¹¹B, n = 6, Table S4), sulfur isotopes (δ³⁴S, n = 5, Table S5), and chlorine isotopes (δ³⁷Cl, n = 18, Table S6).

4. Results and Discussion

Although some hydrochemical and isotopic parameters of the studied brines overlap with ranges reported for marine systems, such overlap alone is not diagnostic of a marine origin [32,33]. In evaporite basins, isotope ratios and elemental compositions are process-sensitive rather than source-exclusive, and similar signatures can be generated through evaporation, burial diagenesis, clay mineral reactions, and residual brine evolution in closed continental settings. In the Mahai Basin, stratigraphic and sedimentological evidence consistently indicates a non-marine depositional environment, precluding a marine source for K-forming fluids. Accordingly, isotopic systems in this study are interpreted as tracers of fluid evolution processes rather than indicators of marine inheritance. This process-oriented approach provides a geologically consistent framework for explaining the observed isotopic characteristics without invoking ancient marine deposits.

4.1. Hydrogen–Oxygen Isotopes and Fluid Evolution

Stable hydrogen and oxygen isotopes are effective tracers of evaporation, halite dissolution, water–rock interaction, and external water mixing, and therefore provide critical constraints on brine evolution, origin, and material sources [34,35]. Different types of brines in the MHB show well-defined clustering in δD-δ¹⁸O space, corresponding to distinct hydrological end members and reflecting the coupled influence of meteoric recharge, lacustrine evaporation, and deep fluid modification (Figure 2; Table S1).

Initial recharge waters, represented by the Yuka River and alluvial fan groundwater, display δD values of −89‰ to −51‰ and δ¹⁸O values of −12.4‰ to −6.3‰ [8,19,21,36,37]. These values plot close to the Global Meteoric Water Lines (GMWL) [38] and Local Meteoric Water Lines (LMWL) [29] and are isotopically lighter than waters from nearby Dachaidan (δD −84.4‰ to −65.8‰) (Figure 1b). Their relatively low d-excess values (mean ~10.7‰) suggest recharge from higher elevations or colder climatic conditions [8]. Elevation effect calculations indicate a dominant recharge altitude of ~3440 m [6], implying that brine precursors primarily originated from high-elevation precipitation and snowmelt from the Qilian, Altyn Tagh, and Saishiteng Mountains [19,39], with minimal evaporation or water–rock interaction during initial recharge.

After entering low-lying lake depressions, surface waters undergo intense evaporation under arid climatic conditions. Lake waters from DZMHL and NLZNL show systematic enrichment in heavy isotopes along the Local Evaporation Line (LEL) [18], with δD values of −61.5‰ to −47.9‰ and δ¹⁸O values of −5.1‰ to −1.3‰. This pattern reflects kinetic fractionation during non-equilibrium evaporation and represents the primary concentration stage of brines [5,18,21].

Intercrystalline and sand-gravel brines derived from infiltrated lake waters exhibit even more pronounced isotopic enrichment (intercrystalline brines: δD −41.2‰ to +5.5‰; δ¹⁸O −2.0‰ to +6.5‰), accompanied by markedly reduced d-excess values. These features indicate prolonged and intensified evaporation relative to surface waters [15,39]. Along the hydrological pathway from alluvial fan recharge to highly mineralized brines, isotopic values increase progressively, suggesting burial and preservation of early evaporative brines under semi-closed conditions with limited water–rock interaction. These brines represent key transitional end members linking surface evaporation systems to deep reservoirs [5,7,21,35]. Deep sand-gravel brines exhibit δD values of −56.9‰ to −17.6‰ and δ¹⁸O values of −5.7‰ to +6.0‰ and generally plot to the right of the LEL. This distribution indicates that they are not derived from direct infiltration of modern surface waters but instead inherit isotopic signatures from intercrystalline brines or paleo-lake brines after evaporation. Some confined brines show positive δ18O values (+1.5‰ to +2.4‰), reflecting isotopic exchange during deep circulation and water–rock interaction. These features suggest downward migration of evaporatively concentrated surface brines and their storage in deep sandy–gravel reservoirs [7,19,25,35,40].

Although evaporation dominates the overall isotopic evolution, anomalous δD-δ¹⁸O signatures in deep brines indicate additional contributions from multi-source mixing and water–rock interaction. Some deep waters (e.g., thermal waters beneath NLZNL) show very low δD values (down to −72.7‰) combined with significantly positive δ18O values (+1.50‰ to +2.40‰), deviating strongly from the LMWL [15,21]. Their atypical δ¹⁸O-Cl-relationships [19] indicate mixing between meteoric waters and deep Ca-Cl-type fluids that have undergone extensive water–rock interaction. The positive δ¹⁸O shifts reflect isotopic exchange with surrounding rocks during deep circulation [6,19,21,25], suggesting late-stage involvement of deep Ca-Cl brines or structurally focused upwelling fluids in the evolution of MHB brines.

4.2. Lithium Isotopes and Fluid–Rock Interaction

In evaporitic lake systems, K is transported and concentrated predominantly as K+ in highly saline brines, and its mineralization is strongly controlled by fluid sources and evolutionary pathways. δ⁷Li is particularly sensitive to low-temperature water–rock interaction and progressive fluid evolution [41,42]; thus, it can provide an effective tracer for the geochemical processes governing K-forming fluids. Mantle-derived or high-temperature deep fluids typically exhibit low δ⁷Li values (−5‰ to 0‰), whereas fluids derived from continental weathering, surface runoff, and evaporative brine systems are characterized by elevated δ⁷Li values (+2‰ to +15‰) [11,43]. In the MHB (Figure 3; Table S2), river waters from the Yuka River display δ⁷Li values of +7‰ to +8‰ [19,21], comparable to those of clay-rich sediments from the DZMHL (+10.96‰) [24]. This consistency indicates that the initial solute load of K-forming fluids was primarily derived from near-surface silicate weathering, with meteoric precipitation and fluvial recharge as the dominant water sources, rather than mantle or magmatic inputs.

Relative to river water end members, alluvial fan region groundwater exhibits significantly higher δ⁷Li values (+20.0‰ to +22.2‰), suggesting that K-forming fluids had already undergone substantial low-temperature water–rock interaction during early shallow migration. During this process, secondary clay minerals preferentially incorporate ⁶Li, progressively enriching residual fluids in ⁷Li [21,44,45]. Along the hydrological gradient from recharge zones toward the basin interior, δ⁷Li values increase from river water to lake water and alluvial fan region groundwater (approximately +10‰ to +26‰), corresponding to a stage when potassium begins to accumulate but major evaporite minerals have not yet precipitated [24,25]. In the basin center, intercrystalline brines and deep confined brines reach the highest δ⁷Li values (+26‰ to +32‰), reflecting highly evolved and high-salinity brines associated with K mineralization stages [21]. The continuous increase in δ⁷Li from recharge areas to the basin center indicates long-term fluid evolution within closed to semi-closed systems. Under low-temperature water–rock interaction conditions, a progressive enrichment of ⁷Li in residual fluids occurs [11,41,42], whereas evaporation and halite precipitation exert only minor fractionation effects on lithium isotopes [19,46]. The gradual isotopic enrichment, coupled with synchronous increases in K⁺, Cl⁻, and total dissolved solids [15,19,24], indicates that K-forming fluids in the MHB evolved through sustained transport and progressive geochemical modification, rather than being dominated by short-lived or episodic fluid inputs. Consequently, the elevated δ⁷Li values of basin brines primarily record cumulative low-temperature water–rock interaction during prolonged migration through Paleogene–Neogene strata, rather than simple evaporative fractionation.

Deep Ca-Cl-type brines and sand–gravel brines have even higher δ⁷Li values, averaging +36‰ and reaching up to +40.2‰, as a result of prolonged fluid–rock interaction at low to moderate temperatures, rather than a surface water source modified solely by geothermal heating [7,19]. The δ⁷Li values of intercrystalline brines and deep confined brines fall between those of river water and Ca-Cl-type brines, suggesting that their isotopic compositions are governed primarily by mixing between these two end members rather than by a single evaporative trajectory [7]. Sand-gravel brines exhibit δ⁷Li values as high as +33.84‰ to +43.82‰, effectively excluding shallow halite dissolution or modern evaporative re-fractionation as dominant controls. Instead, their high δ⁷Li combined with moderate Li concentrations indicates inheritance from deeply evolved interstitial or paleo-lake brines, subsequently transported laterally into alluvial-fan reservoirs while preserving deep-brine isotopic signatures [7,19,21].

Taken together, an empirical positive trend is observed between δ⁷Li values and K⁺ concentrations across different hydrological compartments in the study area. From recharge zones to the basin center, both δ⁷Li and K⁺ concentrations systematically increase, following the sequence of Yuka River waters < DZMHL < NLZNL < solution-mined brines < deep confined brines (Figure 4; Table S2) [21]. However, it is important to emphasize that no direct geochemical correlation or causal control exists between δ⁷Li and K concentrations. Instead, lithium isotopes act as an independent tracer that records the mixing proportions between two contrasting water sources, namely surface-derived waters and deep Ca-Cl waters, whereas the mixing process itself constitutes the primary mechanism governing potassium enrichment in brines [19]. The near-conservative behavior of lithium isotopes during mixing in high-salinity systems further supports this interpretation and suggests that K mineralization in the MHB represents a continuous, multistage evolutionary process. This process initiates with meteoric and fluvial recharge, followed by prolonged evaporation and low-temperature water–rock interaction, and is subsequently modified by mixing with deep Ca-Cl waters. K enrichment is therefore inherited and amplified during fluid evolution within favorable structural and depositional settings, rather than being directly controlled by lithium isotope fractionation.

4.3. Strontium Isotopic Evolution of Brine

The ⁸⁷Sr/⁸⁶Sr ratios of different waters and brines in the MHB range from 0.71106 to 0.71217 (Figure 5; Table S3), distinctly higher than those of mantle-derived materials (0.7035) and modern seawater (0.70939) [46,47], but slightly lower than typical upper continental silicate rocks (0.71190) [48]. This distribution indicates that K-forming fluids in the MHB are dominated by crustal Sr inputs and controlled by multi-end-member mixing rather than mantle-derived contributions [6].

River water from the Yuka River and alluvial fan region groundwater exhibits relatively high ⁸⁷Sr/⁸⁶Sr ratios (0.71150–0.71183) [15,21], although still lower than those of clay-rich sediments from the DZMHL area (0.71239–0.71406) [24]. These values represent a surface recharge end member derived from weathering and leaching of old silicate rocks surrounding the Qilian Mountains. In contrast, deep brines and oilfield waters within the Lenghu anticlinal belt show slightly lower ⁸⁷Sr/⁸⁶Sr ratios (~0.71135) but markedly elevated Sr concentrations, reflecting prolonged water–rock interaction in a deep, relatively closed system, involving dissolution of ancient evaporites and carbonate rocks and associated isotopic exchange [37]. The high ⁸⁷Sr/⁸⁶Sr ratios in riverine end members and the low ⁸⁷Sr/⁸⁶Sr ratios in deep brine end members constitute the fundamental source framework of K-forming fluids in the MHB.

Deep confined brines in the basin are characterized by a narrow range of ⁸⁷Sr/⁸⁶Sr ratios (0.7113–0.7115) [15,21], indicating a relatively uniform strontium isotopic composition. These values fall between those of surface waters and deep brines reported from the Lenghu anticlinal belt [49], yet they do not define discrete isotopic end-members or resolvable mixing proportions. Rather, the limited isotopic variability suggests a homogenized Sr source produced by long-term basin-scale fluid circulation and extensive water–rock interaction. When considered together with hydrochemical characteristics (e.g., elevated K concentrations and depleted Ca and Sr contents) [15] and hydrogen–oxygen isotopic evidence [21], the Sr isotope data indicate that meteoric and fluvial inputs, represented by the Yuka River, constitute the ultimate recharge source of the confined brines. However, these surface-derived signatures have been substantially modified during prolonged subsurface residence, with chemically evolved deep anticlinal brines acting as an additional contributing component rather than a distinct mixing end-member. Notably, the ⁸⁷Sr/⁸⁶Sr ratios of the deep confined brines are generally lower than those of all potential surface waters, including rivers, alluvial-fan groundwaters, lake waters, and intercrystalline brines. This feature precludes simple surface-water mixing as the dominant formation mechanism. Instead, extensive water–rock interaction with surrounding strata, particularly early-formed evaporites, likely played a critical role in lowering ⁸⁷Sr/⁸⁶Sr ratios through mineral dissolution and ion-exchange processes. The wide range of ⁸⁷Sr/⁸⁶Sr ratios reported for halite at different depths (0.71124–0.71177) [21] further supports this interpretation, indicating that deep confined brines represent mixtures of paleo-brines from different geological periods that were progressively homogenized during multistage salt-lake evolution.

Intercrystalline brines are regarded as a key end member in the evaporate evolution sequence of the MHB. Their ⁸⁷Sr/⁸⁶Sr ratios range from 0.71164 to 0.71183, significantly higher than those of local halite and gypsum (0.71126–0.71153) [37] and deep confined brines (0.71130–0.71152), indicating that Sr in intercrystalline brines is not solely derived from evaporite dissolution [15,21,37]. The broad variability of halite ⁸⁷Sr/⁸⁶Sr ratios in MHB (0.71124–0.71177) [37] suggests repeated mixing of paleo-lake brines and ongoing halite dissolution ion exchange processes. Moreover, the similarity of ⁸⁷Sr/86Sr ratios between interstitial brines and clay-rich sediments in the DZMHL area (0.71239–0.71406) [24] indicates a strong control from silicate weathering products sourced from the surrounding Qilian Mountains. Considering their shallow burial depth, Sr isotopic composition, and hydrochemical characteristics, intercrystalline brines in the MHB likely formed through mixing of surface runoff with minor deep Ca-Cl-type paleo-brines, followed by intense evaporative concentration and limited water–rock interaction in a relatively open near-surface environment. This evolutionary pathway resulted in highly evolved brines characterized by elevated ⁸⁷Sr/⁸⁶Sr ratios and Ca-Cl-type chemistry. Their isotopic compositions inherit and amplify the radiogenic Sr signal derived from silicate weathering in the Qilian Mountains, clearly distinguishing them from deep confined brines and highlighting their role as a key end member in the multi-stage evolution of the basin brine system. Overall, Sr isotope evidence demonstrates that K-forming fluids in the MHB are mainly derived from mixing between surface-derived waters and deep structural fluids, effectively excluding large-scale direct mantle involvement in K mineralization.

Based on ⁸⁷Sr/⁸⁶Sr mass balance modeling, intercrystalline brines can be interpreted as typical two-end-member mixtures, with surface water from the Yuka River contributing approximately 58–77% and deep Ca-Cl-type fluids accounting for 23–42% [37]. Spatially, pronounced heterogeneity is observed: brines located near anticlines and fault zones show Sr isotopic compositions and Sr concentrations closer to the deep end member, whereas brines farther from structural belts predominantly reflect river water recharge [37].

4.4. Constraints from Boron, Sulfur, and Chlorine Isotopes

Boron isotopes provide critical constraints on fluid sources and water–rock interaction processes in the MHB. Brines associated with anticlinal structures exhibit highly positive δ¹¹B values (+26‰ to +37‰), similar to those of BLMHL (+31.4‰) and NLZNL (+23.7‰ to +24.5‰), whereas Yuka River water (−5.7‰), DZMHL (−4.0‰), and the Nanbaxian salt pits (−12‰ to −24‰) show low to extremely low δ¹¹B values (Table S4) [22,23]. This wide isotopic range indicates the involvement of at least three distinct boron end members. The close agreement between δ¹¹B values of DZMHL and the Yuka River suggests that low δ¹¹B boron derived from continental weathering is continuously supplied to lakes and shallow brine systems through surface runoff and shallow infiltration. In contrast, the high δ¹¹B values of NLZNL and BLMHL overlap with those of deep paleo-brines from the Lenghu anticlinal belt [18,22,23], reflecting prolonged water–rock interaction and brine evolution under deep, closed conditions. These deep fluids play a key role in the co-enrichment of K. Extremely low δ¹¹B values in Nanbaxian brines and solution-phase sediments (−34.71‰ to −6.14‰) [22], comparable to those of solid borates in the Dachaidan Lake area, indicate a dominant contribution from weathering and leaching of rocks in ultra-high-pressure metamorphic belts [22]. Overall, the evolution of ore-forming fluids in the MHB is governed by coupling between low δ¹¹B surface runoff, high δ¹¹B deep paleo-brines, and structurally controlled fluid inputs. This pattern contrasts sharply with the central QB, where mineralization is dominated by low δ¹¹B volcanic geothermal fluids [18]. The markedly high δ¹¹B values in western MHB brines (up to +31.4‰) provide clear evidence for the ascent of deep paleo-brines along structural pathways, representing a primary driving force for depth-dependent enrichment of K and the formation of highly mineralized brines and K deposits.

Sulfur isotopes further constrain redox conditions and external inputs during brine evolution. Deep confined brines in the MHB show δ³⁴S values of 12.02–16.57‰ [25], higher than those of modern shallow salt lakes such as Lop Nor and Qarhan (6.66–19.9‰) [50,51], but lower than those of deep gravel-hosted brines and Paleogene–Neogene solid sulfates in western Qaidam (20.84–39.2‰) (Table S5) [18,52]. These intermediate δ³⁴S values indicate a balance between enrichment driven by bacterial sulfate reduction and water–rock interaction in semi-closed deep reservoirs and dilution caused by leaching of sulfate-bearing strata in the northern basin margin and by meteoric recharge (8.65‰) [25]. Thus, sulfur isotopes reveal a unique evolutionary regime in which K-forming fluids evolved predominantly under closed conditions but remained subject to limited external freshwater input.

Chlorine isotopes provide decisive evidence for the origin of sand-gravel brines. Their δ³⁷Cl values range from −0.26‰ to +0.25‰, significantly lower than those of halite (+0.41‰ to +0.74‰) in MHB (Table S6) [7]. The Na/Cl molar ratios of sand-gravel brines (0.80–0.82) [7] are far below the theoretical value of 1 expected for simple halite dissolution. These independent geochemical indicators rule out Neogene halite dissolution as the primary source of sand-gravel brines [53]. During progressive evaporation and halite precipitation in evaporitic lake systems, ³⁷Cl is preferentially incorporated into solid halite, leading to progressively lower δ³⁷Cl values in residual brines [54]. The depleted δ³⁷Cl signatures of sand-gravel brines are therefore consistent with their interpretation as highly evolved residual fluids. By contrast, freshwater sources such as precipitation and river water typically exhibit positive δ³⁷Cl values (+0.74‰ to +4.48‰) in MHB [54], clearly distinct from freshwater sources and sand-gravel brines in isotopic space, confirming their deep origin. The δ³⁷Cl values of sand-gravel brines closely overlap with those of intercrystalline brines (DZMHL) or subsurface brine, indicating inheritance of geochemical signatures from deeply evolved brines and a common genetic origin [7]. Integrating δ³⁷Cl and Na/Cl systematics allows reconstruction of the evolutionary pathway of sand-gravel brines: meteoric waters with high δ³⁷Cl underwent strong evaporation and halite precipitation (Table S6), during which ³⁷Cl was preferentially removed into solid salts, producing low δ³⁷Cl deep interstitial brines; these highly evolved fluids subsequently migrated laterally and were finally stored within gravel reservoirs [7]. This isotopic evidence conclusively excludes a shallow dissolution origin and identifies sand-gravel brines as residual products of continuous deep brine evolution with clear metallogenic significance.

4.5. Cross-Isotope Constraints on Fluid End-Members and Evolutionary Processes

Integrated constraints from multiple isotope systems (δD-δ¹⁸O, δ⁷Li, ⁸⁷Sr/⁸⁶Sr, δ¹¹B, δ³⁴S, and δ³⁷Cl) allow the evolution of K-forming fluids in the MHB to be evaluated within a unified geochemical framework, rather than through separate isotope-by-isotope interpretations. Cross-isotope synthesis indicates that fluid evolution in the basin can be reasonably described by interactions among three conceptual end-members: (1) a meteoric surface-recharge end-member, (2) a highly evolved intercrystalline brine end-member, and (3) a deep-seated Ca-Cl waters end-member. These end-members represent evolutionary stages and dominant processes rather than discrete, chemically fixed fluids.

During the initial concentration stage, hydrogen–oxygen isotopes display systematic enrichment trends along the Local Evaporation Line (LEL), indicating strong evaporative control under arid climatic conditions [6,8,21,25,29,35,40]. This behavior is closely aligned with a moderate increase in δ⁷Li values [19,24], suggesting that while evaporation dominated the hydrological evolution, low-temperature water–rock interaction had already begun to influence solute compositions. In this stage, δD-δ¹⁸O isotopes primarily record physical water loss, whereas lithium isotopes are sensitive to early-stage fluid–mineral interaction. Radiogenic ⁸⁷Sr/⁸⁶Sr ratios (0.7115–0.7118) observed in shallow waters and early brines further support a meteoric recharge origin, reflecting interaction with silicate-rich lithologies of the Qilian Mountains [6,15]. The consistent behavior of H-O, Li, and Sr isotopes at this stage indicates that surface-derived waters underwent basin-scale evaporation and initial geochemical modification prior to deeper circulation.

As fluids evolved and migrated into deeper parts of the basin, their isotopic compositions progressively deviated from simple evaporation trajectories. Deep confined brines and sand-gravel brines exhibit a pronounced positive δ¹⁸O shift [7,25,35], accompanied by elevated δ⁷Li values (up to ~+40‰) [7,21] and high δ¹¹B values (+26‰ to +37‰) [18,22]. When considered jointly, these signals indicate prolonged residence times and intense fluid–rock interaction within deep reservoirs, rather than continued surface evaporation. Although some isotope signatures may appear contradictory when considered individually, for example, depleted δ³⁷Cl values in sand-gravel brines compared with positive δ³⁷Cl values in surface waters, these differences can be reconciled through fractionation processes associated with halite precipitation. Preferential incorporation of ³⁷Cl into halite during late-stage evaporation results in isotopically light residual brine [54]. Consequently, low δ³⁷Cl values represent a diagnostic feature of highly evolved residual brines, rather than evidence against deep-fluid involvement. In this context, chlorine isotopes provide a complementary constraint on the advanced evolutionary state of deep fluids that are critical for K enrichment.

Cross-isotope consistency is further demonstrated by the combined behavior of strontium and boron isotopes. Variations in ⁸⁷Sr/⁸⁶Sr ratios suggest that intercrystalline brines reflect mixing between surface-derived fluids and deep Ca-Cl-type fluids, with relative contributions varying spatially and structurally across the basin [15,21,37]. These interpretations are independently supported by δ¹¹B systematics [18], which indicate that potassium co-enrichment is closely associated with the upward migration of boron-rich deep paleo-brines along structural pathways. Rather than relying on a single isotope system, the convergence of Sr and B isotope evidence strengthens the inference that deep-fluid modification played a key role in the late-stage evolution of K-forming fluids. Importantly, the inferred mixing relationships should be viewed as semi-quantitative constraints that capture first-order processes rather than exact mass balance solutions.

By integrating multiple isotope systems across successive evolutionary stages, the apparent complexity and partial decoupling among isotopic signals can be reconciled within a coherent framework. K mineralization in the MHB is best understood as the cumulative outcome of high-altitude meteoric recharge, basin-scale evaporation, and subsequent deep fluid modification. Each isotope system records a different aspect of this evolution, and their combined behavior provides mutually reinforcing constraints on the sources, pathways, and transformation processes of K-forming fluids. This integrated cross-isotope framework establishes a complete metallogenic chain of “surface runoff recharge, intense evaporative fractionation, deep fluid mixing, lateral migration and storage”, providing a robust basis for developing a tectonic-fluid-mineralization coupling model for the MHB, and for guiding exploration of concealed gravel-type K deposits in arid inland basins.

5. Conclusions

Multi-stable isotope analysis (δD-δ¹⁸O, δ⁷Li, ⁸⁷Sr/⁸⁶Sr, δ¹¹B, δ³⁴S, and δ³⁷Cl) indicates that K mineralizing fluids in the MHB represent a multi-source, long-lived, and evolutionarily complex system. δD-δ¹⁸O reveals that the initial water was mainly derived from atmospheric precipitation and snowmelt in the Qilian Mountains and surrounding highlands, delivered via river runoff, and subsequently concentrated through prolonged evaporation under arid and semi-closed basin conditions, forming a continuous evolution from river water to lake water to brine, with late-stage δ¹⁸O enrichment reflecting mixing with deep circulating fluids and water–rock isotope exchange. δ⁷Li systematically increases along the evolution sequence and correlates with salinity and K content, indicating long-term low-temperature water–rock interactions and selective adsorption of ⁶Li by secondary clay minerals, while deep Ca-Cl brines with exceptionally high δ⁷Li represent highly evolved endmembers from deep cycling. Elevated ⁸⁷Sr/⁸⁶Sr values exclude significant mantle contributions, confirming that solutes mainly originate from weathering of Qilian silicate rocks, with interstitial brines acting as key intermediate endmembers between surface and deep fluids. δ¹¹B, δ³⁴S, and δ³⁷Cl further indicate that deep fluids ascending along anticlines and fault zones drive K enrichment, whereas sand-gravel brines represent highly evolved residual fluids inheriting paleo-lake evaporation signatures. These combined multi-isotope constraints establish a comprehensive evolutionary model for MHB K mineralization, in which surface runoff supplies the initial water, followed by intensive evaporation and low-temperature water–rock interaction, mixing with deep Ca-Cl paleo-brines along structural conduits, lateral transport and interlayer entrapment, ultimately leading to K precipitation.