Geochemical Characteristics and Origin of Heavy Metals and Dispersed Elements in Qarhan Salt Lake Brine

Abstract

This study investigated the distribution and source of heavy metals and dispersed elements in the high-salinity brine of Qarhan Salt Lake. The brine with an average total dissolved solid content of 332.22 g/L, dominated by Cl⁻ (216.41 g/L) and Mg²⁺ (44.76 g/L), indicated strong evaporation and dolomite dissolution. As (6.57 ± 3.59 μg/L) and Hg (0.48 ± 0.14 μg/L) showed uniform distribution while Li (69.66 mg/L), B₂O₃ (317.80 mg/L), and Zn (5.69 mg/L) were highly enriched, highlighting the resource potential and geochemical complexity. Correlation analysis revealed that water–rock interaction played a key role in element differentiation, with Sr and Ca²⁺/Cl⁻ showing strong positive correlations (r = 0.693/0.768), reflecting isomorphic substitution and dissolution. Meanwhile, Na⁺ and Mg²⁺/Ca²⁺ showed negative correlations (r = −0.732/−0.889), suggesting cation exchange and gypsum precipitation. The self-organizing map yielded four clusters of elements and positive matrix factorization model identified four sources; the elements in the Salt Lake brine mainly came from the river water supply, weathering and leaching of minerals, and dissolution of salt-bearing layers and were locally influenced by human activities. The research provided valuable insights for future sustainable development and the environmental protection of the region.

1. Introduction

Salt lakes, as important natural resource enrichment areas, are not only important sources of mineral resources but also natural laboratories for studying geochemical processes. The unique arid climate and geological background of the Qarhan Salt Lake result in significant regional characteristics in the distribution and occurrence of constant ions, heavy metals, and dispersed elements in the brine [1]. In high-salt brine systems, the transportation and transformation of heavy metals and dispersed elements are regulated by various environmental parameters such as electronic conductivity (EC), pH, dissolved oxygen (DO), etc. These factors drive the adsorption desorption equilibrium and morphological transformation of elements by changing the surface charge characteristics [2]. In addition, the high-mineralization characteristics and ion composition of brine not only reflect strong evaporation and concentration but also provide important clues for analyzing the geochemical evolution of salt lakes.

The presence of hazardous pollutants in water sources has become a major environmental challenge, attracting significant attention in recent years [3]. Heavy metals are toxic and persistent and can migrate and accumulate with the environment, causing ecological damage and health risks [4,5]. Brine contains a high concentration of valuable elements such as Li and B, indicating significant strategic resource potential. At the same time, the presence of heavy metals and the complexity of its chemical composition may pose potential environmental risks if not properly managed. Heavy metal pollution in salt lakes can cause multiple hazards to ecosystems and resource utilization. Heavy metals such as cadmium (Cd), lead (Pb), and mercury (Hg) can inhibit the activity of microorganisms in salt lakes, disrupt the sulfur and nitrogen cycles, and lead to a decline in biodiversity [6]. Heavy metals, after being enriched by organisms such as brine worms and algae, may be transmitted along the food chain, threatening waterbirds and surrounding residents [7]. In addition, heavy metals can reduce the purification efficiency of potash, lithium, and other resources in salt lakes and increase development costs and may also contaminate surrounding soil and farmland through underground water migration or wind erosion diffusion [8]. Heavy metal contamination not only threatens the ecological balance of salt lakes but also poses significant risks to resource development and regional environmental security. Therefore, systematic research on the chemical parameters, ion composition, and concentrations of trace metals in brine not only helps reveal its geochemical processes but also provides a scientific basis for resource development and environmental protection. It is crucial for protecting high-salt fragile ecosystems and preventing ecological and health risks. The research aims to systematically analyze the brine samples from the Dabuxun section in the demonstration area of potassium dissolution mining, clarify the regulatory mechanism of its main chemical indicators (such as temperature, pH, DO, and EC) on heavy metal migration, reveal the composition characteristics of constant ions and their evaporation and concentration effects, and analyze the content of heavy metals and dispersed elements to provide a scientific basis for the efficient resource utilization of the Qarhan Salt Lake.

Multi-technology combination and model simulation methods were used to study the geochemistry characteristics and source identification of heavy metals and dispersed elements in water and sediments. The inductively coupled plasma mass spectrometer (ICP-MS) and X-ray fluorescence spectrometer (XRF) were used to perform a quantitative analysis of the element contents. This was combined with enrichment factors and the geoaccumulation index to distinguish the natural geological background from anthropogenic pollution [9]. We used isotope ratios such as ⁸⁷Sr/⁸⁶Sr to accurately trace sources of pollutants (such as industrial emissions, agricultural inputs, or primary mineral dissolution) [10]. The speciation of heavy metals in sediments was studied by scanning electron microscopy–energy-dispersive spectroscopy (SEM-EDS) [11]. In addition, statistical models such as principal component analysis (PCA) and positive matrix factorization (PMF) can quantify the contribution of multi-source pollution [12].

2. Materials and Methods

2.1. Research Area

The Qarhan Salt Lake is located on the Qinghai–Tibetan Plateau and has an average annual temperature of 5.2 °C [13]. It is composed of Dabuxun, Nanhuobuxun, Beihuobuxun, and Senei Salt Lake, forming a vast salt-lake complex system. Its east–west extension exceeds 160 km, and its north–south width ranges from 20 to 40 km [14]. After long-term high-concentration brine evaporation and crystallization, the Qarhan Salt Lake has nurtured a super-large soluble potassium magnesium salt deposit that is rare in the world [15]. This research area is situated in the Dabuxun area of Qarhan Salt Lake, on the north bank of Dabuxun Lake (Figure 1). Hydrological observation holes are generally arranged on different elevations, geological units, or hydrological gradients of dry salt flats to reflect the dynamic changes of groundwater. Exploration pits are usually located in the surface salt crust, evaporite rock, or potential aquifer area, which are used for visual observation of stratigraphic structure and salt deposition. Water replenishment channel is usually located on the edges of dry salt flats or near Dabuxun Salt Lake and feeds the salt lake region through artificial or natural runoff.

2.2. Field Sample Collection and Testing

This research collected underground brines from the demonstration region of potassium dissolution mining in the Dabuxun section of the Qarhan Salt Lake. In July 2022, 15 brines (0–20 cm) were collected from 9 hydrological observation holes, 3 exploration pits, 1 water replenishment, 1 brine channel, and Dabuxun Lake in the research area (Figure 1). We investigated the hydrochemical characteristics of underground brine during liquefaction mining in the area. The sampling process strictly followed the specifications of the “Water quality—Guidance on sampling techniques” (HJ 494-2009) standard [16], and a sampler with an inner wall washed with 5% pure HNO₃ was used for brine collection. The contents of K⁺, Na⁺, Mg²⁺, and Ca²⁺ constant cations were measured using an inductively coupled plasma optical emission spectrometer (ICAP6300, Thermo Fisher, Waltham, MA, USA) and the contents of anions (Cl⁻, SO₄²⁻, HCO₃⁻, CO₃²⁻) were determined using an ion chromatograph (ICS1100, Thermo Fisher, Waltham, MA, USA). The contents of As and Hg were measured via atomic fluorescence spectrometry (AFS-8330, Haiguang Instrument, Beijing, China) and the contents of Pb, Cd, Cu, Ni, Zn, Cr, Li, Sr, B₂O₃, Br, Rb, and Cs were measured by inductively coupled plasma mass spectrometry (Agilent 7900, Agilent, Santa Clara, CA, USA). While measuring elements in brine using ICP-MS, the mineralization should be below 2 g/L. Excessively high mineralization will cause the salt to block the sampling cone. One must simultaneously insert 10% parallel samples and standard substances for quality control and control the recovery rate of each element within 85–115% (RSD < 5%) to ensure the accuracy of trace element detection in high-salt brine.

2.3. Assessment Methods

2.3.1. Self-Organizing Map (SOM)

In this study, the SOM toolbox in MATLAB was used to quantify the spatial clustering characteristics of elements in brine. SOM is able to map complex multidimensional data dimensionality reduction to two-dimensional feature planes through competitive learning mechanisms [17]. The model’s accuracy was evaluated through quantization error (QE) and terrain error (TE). As an important branch of artificial neural networks, SOM demonstrates superior performance in high accuracy, visual representation, and minimal manual intervention [18].

2.3.2. Positive Matrix Factorization (PMF) Models

The PMF model has significant application value in the field of multi-source pollution tracing in aquatic systems [19]. The algorithm ensures the physical interpretability of the interpretation results by simultaneously processing missing data, error estimation, and non-negative constraints and has thus been established as a standard method for resolving element sources in environmental research [20].

$$X_{ij}={\sum }_{k=1}^p{g}_{ik}{f}_{kj}+e_{ij}$$

(1)

$$\mathrm{Q}={\sum }_{i=1}^n{\sum }_{j=1}^m{({\displaystyle \frac{e_{ij}}{u_{ij}}})}^2$$

(2)

$$U_{ij}=\left\{{\displaystyle \frac{5}{6}}MDL,X_{ij}\le MDL|\sqrt{{(P·C)}^2+{\left(MDL\right)}^2, }{X}_{ij}>MDL\right\}$$

(3)

Here, the content of element j is represented as $X_{ij}$; $f_{kj}$ is the characteristic value of pollutant source k on the content of element j; $e_{ij }$ is the residual matrix; $u_{ij}$ is the uncertainty of element j; C is the measured element content.

2.3.3. Statistical Analysis

The violin plots, piper trilinear diagram, and correlation analysis were generated in Origin 2023b. The map of the research area was drawn using ArcMap 10.2. MATLAB 2012a was used for SOM of elements in brines and cluster distribution combined with k-means. We performed quantitative source apportionment using EPA PMF 5.0 software.

3. Results and Discussion

3.1. Main Chemical Indicators of Salt Lake Brine

The form of heavy metals in natural water is significantly regulated by the solid–liquid distribution mechanism, and the proportion of water-soluble states is generally low (usually <5%), mainly fixed on the solid through complexation with suspended particle surfaces or electrostatic adsorption [21]. Environmental factors such as the pH, DO, and EC drive the adsorption and desorption equilibrium and morphological transformation of heavy metals by altering the surface charge characteristics and oxidative potential of particulate matter [22]. This study used a multi-parameter water quality sensor to systematically measure key parameters such as the temperature (T), dissolved oxygen (DO), pH, and oxidation–reduction potential (ORP) of the brine (Figure 2), providing quantitative environmental constraints for analyzing the interface process of heavy metal migration in high-salt brine systems.

Water temperature is a key parameter that characterizes the state of water environment and regulates biochemical reaction processes and its dynamic changes are often closely related to climatic conditions [23]. The temperature of brine in the research area ranged from 22.02 to 23.71 °C (average 22.92 °C), reflecting the relatively stable environment of the typical arid-to-semi-arid climate on the Qinghai–Tibetan Plateau in summer. The pH value ranged from 6.32 to 7.55 (mean 7.18), indicating that the brine was weak alkalinity. Dissolved oxygen was the core indicator of redox state; its concentration was 2.18 ± 0.61 mg/L. Together with electronic conductivity (479.03 ± 36.82 mS/cm), total dissolved solids (335.32 ± 25.77 g/L), and salinity (274.56 ± 20.40 ppt), it revealed the high-mineralization characteristics of brine.

The oxidation–reduction potential of brine was −292.70~79.10 mV, with a mean value of −131.67 mV. The spatial differentiation revealed the complex oxidation–reduction characteristics of the brine system. The negative zone (ORP < 0 mV, accounting for 93%) indicated that most of the brine was in a weak-oxygen/oxygen-deficient environment, which may have been related to the limited oxygen diffusion caused by the deep brine being sealed in low-permeability sedimentary layers. And positive values (ORP > 0 mV) were distributed in the active water exchange zone, suggesting a correlation with surface runoff input or oxidative water mixing caused by artificial brine extraction [22]. This ORP distribution (negative skewness; skewness coefficient was −0.68) was not only controlled by brine retention time and supply mode but may also have reflected the regulation of electron transfer chains by geochemical processes such as sulfate reduction and organic matter degradation.

3.2. Characteristics of Major Constant Ionic Composition in Salt Lake Brine

The ion concentration of brine in the research area showed that Cl⁻ was the main dominant component, with a concentration of 216.41 ± 23.15 g/L, significantly higher than those of other ions. The average concentrations of Na⁺, Mg²⁺, and K⁺ were 46.49 ± 21.37, 44.76 ± 17.81, and 17.47 ± 15.09 g/L, respectively. The contents of SO₄²⁻, Ca²⁺, and CO₃²⁻ were relatively low, with values of 5.02 ± 1.64, 1.51 ± 0.64, and 0.38 ± 0.16 g/L, respectively. The concentration of HCO₃⁻ was the lowest, with 0.14 ± 0.05 g/L. Changing the content of constant ions can affect the combination and distribution of ions in water, leading to the reprecipitation or precipitation of heavy metals and causing the secondary pollution of water [24].

The Piper trilinear diagram can analyze the hydrochemical characteristics of different types of water. It consists of a diamond and a pair of equilateral triangles arranged to form a large triangle, with the two triangles representing anions and cations, respectively, connected by the diamond in the middle. Specifically, different types of water are classified based on the interrelationships between the eight main ion components in water.

The sampling points corresponding to the triangular area in the lower left were mostly distributed on the right side, close to the Mg²⁺, followed by the Na⁺ + K⁺ area (Figure 3). Cations were dominated by Mg²⁺ (73.8%) and Na⁺ (20.2%), with weak contributions from K⁺ (4.5%) and Ca²⁺ (1.5%). This distribution suggested that the enrichment of Mg²⁺ may be related to the relative accumulation of Mg²⁺ in residual brine after early Ca²⁺ precipitation through carbonate in the dissolution or evaporation sequence of dolomite [25]. Although the absolute concentration of Na⁺ was high (average value was 46.49 g/L), its monovalent characteristic resulted in a lower equivalent ratio than that of Mg²⁺, suggesting that NaCl may not have fully saturated and precipitated. The sampling points corresponding to the triangular area in the lower-right corner were mostly located near the Cl⁻ axis and had a high percentage of the milligram equivalent. Cl⁻ accounted for more than 90% of the anions in all brine samples, close to 100%, far exceeding HCO₃⁻ + CO₃²⁻ and SO₄²⁻, indicating that Cl⁻ was the main anion. The brine sample points in the diamond shaped area were concentrated in the upper-right corner and were distributed in the transition zone between Mg-Cl and Na-Cl types, leaning towards the Mg-Cl, revealing that the brine chemical type was of the MgCl₂-NaCl mixed type.

Natural brine undergoes evaporation and concentration in the salt field, leading to a marked elevation in Mg²⁺, Na⁺, K⁺, and Cl⁻ concentrations. The ion composition of the brine exhibited typical high salinity characteristics, with Cl⁻ occupying an absolute dominant position, indicating that the brine system had undergone strong evaporation and concentration [26]. River and spring water are the two primary recharge sources for the Qarhan Salt Lake. The Golmud River, situated along the south of the Qarhan Salt Lake, is a major inflow source to the Qarhan Salt Lake [27]. The river inflowing the Qarhan Salt Lake is concentrated in Na⁺, Cl⁻, and less Ca²⁺ [28]. The content of CO₃²⁻ and HCO₃⁻ in the Qarhan Salt Lake was significantly low, mainly controlled by the physical and chemical conditions and the process of salt formation. High-salinity brine (dominated by Cl⁻) inhibited carbonate dissolution equilibrium through the “ionic strength effect”, leading to a decrease in CO₃²⁻ activity [29]. The strong evaporation and enrichment promoted the preferential precipitation of calcite and dolomite, continuously consuming the carbonate components in the water [30]. In addition, although the weakly alkaline environment of the salt lake was conducive to the conversion of HCO₃⁻ to CO₃²⁻, the combination of Ca²⁺ and CO₃²⁻ formed CaCO₃, further limiting the occurrence of dissolved carbonate ions.

In addition, a comparative analysis was conducted on the brine sample data in the investigated region. The results indicated that the ion composition of this study area showed both common characteristics and significant differences compared to other salt lakes (

Table 1. Comparison of concentrated brine ion content with other salt lakes (g/L).

Ions	This Area	Kunteyi Salt Lake	Mahai Salt Lake	Xitaijinar Salt Lake	Lop Nur Salt Lake	Zabuye Salt Lake
Na+	46.49	97.10	86.69	86.88	98.00	135.42
K+	17.47	11.80	2.44	7.11	8.90	36.56
Mg2+	44.76	20.38	7.17	23.81	21.40	0.00
Ca2+	1.51	2.60	4.04	/	0.08	0.00
Cl−	216.41	178.97	160.72	186.38	172.00	166.04
SO42−	5.02	54.82	2.48	31.84	56.00	44.78
References		[31]	[33]	[34]	[32]	[35]

Note: “/” indicated data not provided in the reference.

). All salt lakes were mainly composed of Cl⁻ and Na⁺, reflecting the strong evaporation and concentration under arid environments, which conformed to the typical ion enrichment rule of salt lakes. However, this study area exhibited the unique characteristics of a high concentration of Cl⁻ (216.41 g/L), high Mg²⁺ (44.76 g/L), and moderate K⁺ (17.47 g/L). The concentration of Cl⁻ was the highest among all compared salt lakes, which may have been related to deep brine supply or higher-level evaporation. The concentration of Mg²⁺ was higher than that of the Kunteyi Salt Lake (20.38 g/L) [31] or Lop Nur Salt Lake (21.40 g/L) [32], suggesting that dolomite filtration and magnesium precipitation were inhibitory conditions. The low values of SO₄²⁻ (5.02 g/L) in this study area and Mahai Salt Lake (2.48 g/L) [33] reflected the completion of sulfate reduction or early salt precipitation. In addition, the difference in Na⁺/Cl⁻ revealed fractionation characteristics, with a ratio of 0.21 in this study area, significantly lower than that of the Kunteyi Salt Lake (0.54) [31] and Xitaijinar Salt Lake (0.47) [33], indicating a higher stage of evaporation and concentration or the influence of deep brine supply. In summary, the differences in ion combinations among different salt lakes were controlled by structural sealing, water supply sources, and precipitation sequences. The characteristics of high chlorine, potassium, and magnesium in this research area were favorable for the enrichment of rare elements, particularly B and Li, and had greater potential for potassium development.

3.3. Distribution Characteristics of Heavy Metals and Dispersed Elements in Brines

The distributions of As (6.57 ± 3.59 μg/L) and Hg (0.48 ± 0.14 μg/L) were relatively uniform. The mean values of Pb and Cr were 36.76 and 29.03 μg/L, the lowest values were all below the detection limit and were not detected, and the highest values were 111.15 and 40.75 μg/L. The mean values of Cu, Ni, and Zn were 714.28, 318.62, and 5697.56 μg/L. All were significantly enriched, with the highest values reaching 1643.39, 1469.87, and 10,032.06 μg/L, respectively (Figure 4). Geothermal fluids are rich in Ni, Cu, and Zn and flow as upwelling occurs along deep faults in the deep crust, which then mix with shallow groundwater and eventually flow into an enclosed lake through surface runoff or spring water to form a heavy-metal-rich salt lake [36]. Through the comparative analysis of heavy metal contents in several salt lakes, the concentration of Cr in the Qarhan Salt Lake was lower than that of the Keke Salt Lake (38.83 μg/L) and significantly higher than that of the Zabuye Salt Lake (21.70 μg/L), Xiao Qaidam Lake (4.98 μg/L), and Eren Salt Lake (13.40 μg/L) [37]. The high levels of Zn, Ni, and Cu in the Qarhan Salt Lake suggested the possibility of specific geochemistry processes or human influence. Cd values in all the samples and some sampling points of Cr and Pb were below the detection limit and thus not detected, reflecting the low background characteristics of the region. Due to the complex brine system and the relatively high concentrations of heavy metals in salt lakes, brine was not suitable for the “Environmental Quality Standards for Surface Water” (GB3838-2002) [38]. Therefore, there was no applicable evaluation method for brine.

The distributions of elements in the brine were influenced by multiple mechanisms. The Qaidam Basin had experienced multiple hydrothermal activities and volcanic eruptions in geological history, with hydrothermal fluids often rich in Cu, Ni, and Zn [39]. Hydrothermal activities led to the formation and enrichment of sulfide minerals, which are dissolved in subsequent weathering and water–rock interactions, releasing elements into brine [40]. Moreover, through the evaporative enrichment of salt lake brine, elements such as Cu and Zn form complexes or soluble salts with Cl⁻, Br⁻, and B⁻, remaining in the brine. The Zn content was significantly higher than that of typical salt lakes, which may have been driven by the anthropogenic processes [41]. Pb and Cr have low solubility under high-salinity-value conditions and are prone to precipitation or adsorption on particulate matter. Therefore, the migration ability of Pb and Cr in brine is weak, resulting in lower concentrations. Considering that the Qarhan Salt Lake is a production and processing factory, the abnormally high values of Pb and Cr may have been due to human activities [42]. In general, Pb and Zn are tracers of traffic activity, and there are many vehicles transporting products every day in the Qarhan Salt Lake mining area, so traffic activity likely contributes to the elevated Pb and Zn contents at local sampling sites [41]. Cd was not detected; this was consistent with the regional geochemical background. The research results revealed the dual attributes of the “resource development-environmental risk” of salt lake brine, providing a key basis for sustainable utilization.

The dispersed elements in the brine of Qarhan Salt Lake exhibited significant differentiation characteristics. The abundance of Li, Sr, and B₂O₃ were relatively high, with average values of 69.66, 42.39, and 317.80 mg/L, ranging from 40.10 to 95.35, 29.62 to 94.31, and 220.90 to 392.03 mg/L, respectively. The richness of Li and B in the brines basically followed “hot spring supply–river carrying transport–tail salt lake enrichment and mineralization” [43,44]. The concentration range of Br was the largest, with 46.92 ± 38.97 mg/L. In contrast, the abundances of Rb and Cs were significantly lower with 3.42 ± 2.75 and 0.097 ± 0.095 mg/L, respectively.

The distribution characteristics of dispersed elements in the Qarhan Salt Lake revealed multiple control mechanisms for the geochemical evolution of regional brine. Previous studies had shown that the East Kunlun Mountains were rich in Li and B hot springs, which can supply Li and B to the central salt lakes (including Yiliping, Xitaijinar Salt Lake, Dongtaijinar Salt Lake, and the Beiletan section of the Qarhan Salt Lake) [45,46]. The enrichment of Sr may be controlled by the weathering of sulfate minerals such as celestite while the violent fluctuations of Br may be related to the different salt mining areas [14]. In the four sections of the Qarhan Salt Lake, the contents of Br in the Beiletan and Dabuxun Salt Lakes were supplied through riverine inputs and concentrated via evaporation; the Qarhan and Huobuxun Salt Lakes were supplied by the Ca-rich spring water. The elevated Br concentrations in deep brine reservoirs primarily originated from water–rock reactions and the release of Br from sediments and organic substances [47]. The glacier snow on the Kunlun Mountains and the water in the Golmud River and Nalenggele River had low Br levels; after evaporation and concentration, part of the Br source was supplied to the Qarhan Salt Lake [48]. The Qarhan Salt Lake mainly relied on surrounding river supply to flow into it. The concentration of Rb in the supply rivers was low, and these rivers were mostly seasonal small rivers. Moreover, clay minerals such as illite and montmorillonite and organic matter were widely developed in floodplains and alluvial fans around the lake basin, which had strong adsorption capacity for Rb and Cs [49]. The elements that were already extremely scarce in the river water were heavily lost in the clay around the lake basin, making it difficult for them to accumulate in the lake basin, resulting in them having the lowest concentrations [50].

The spatial patterns of heavy metals and dispersed elements in the brine of Qarhan Salt Lake exhibited significant synergistic differentiation characteristics. The high-abundance-element group was dominated by Li and B₂O₃, as well as by heavy metals such as Zn and Cu. The high values of Zn and Cu related to the weathering and strengthening of zinc rich bedrock while the severe fluctuations of Zn and Cu suggested the influence of local human activities [51]. The enrichment of Li and B₂O₃ highlighted their strategic resource potential, which can support the development of the new energy industry chain, while one must beware the ecological risks associated with high levels of Zn. Future development needs to balance efficient resource extraction and the dynamic monitoring of heavy metals to achieve a balance between ecological protection and resource utilization.

3.4. Correlation Analysis Between Heavy Metal and Constant Elements

Multivariate statistical analysis revealed complex water–rock interactions in the research system (Figure 5). At a high significance level of p ≤ 0.01, Na⁺ and SO₄²⁻ showed a strong positive correlation (r = 0.843), indicating that the dissolution of sodium sulfate minerals may have been the main source of ions. The significant positive correlation between Ca²⁺ and Cl⁻ (r = 0.829) highlighted the control of brine composition by evaporation concentration. The geochemical behavior of Li exhibited multidimensional correlation characteristics. The significant positive correlation with B₂O₃ (r = 0.736) reflected that these two substances with similar sources, both come from hot spring recharge [46], while the positive correlation with HCO₃⁻ and CO₃²⁻ (r = 0.832/0.721) suggested their stable occurrence in carbonate systems. The significant positive correlation between Sr and Ca²⁺ and Cl⁻ (r = 0.693/0.768) was consistent with the isomorphic substitution mechanism during co precipitation. Sr²⁺ can enter CaSO₄ to replace Ca²⁺, and during the dissolution and precipitation of CaSO₄, the concentration of Ca²⁺ increases at the same rate as that of Sr²⁺.

At the significance level of p ≤ 0.05, the positive correlation between K⁺ and Ca²⁺ (r = 0.629) and the correlation with Cl⁻ (r = 0.732) reflected that evaporite mineral dissolution is a dominant process in the salt lake system. Sylvite and carnallite are likely the primary sources of K⁺ and Cl⁻. Ca²⁺ is being supplied from carbonate or sulfate minerals, possibly from the same source rock weathering or from adjacent brine systems. The three ions may originate from the same geological units, such as from evaporate deposits or altered sedimentary rocks. This pattern is typical in closed-basin salt lakes like the Qarhan Salt Lake, where chemical processes are dominated by mineral dissolution and evaporation. The positive correlation between Mg²⁺ and Ca²⁺ and Cl⁻ (r = 0.550/0.611) indicated that the dissolutions of magnesium salt including of dolomite or carnallite were the main sources of Mg²⁺, Ca²⁺, and Cl⁻ in the brine [52]. The negative-correlation matrix of Li was particularly unique, with negative correlations with Pb and Zn (r = −0.618/−0.554) suggesting competitive adsorption effects, while negative correlations with K⁺ (r = −0.575) may have reflected the selective adsorption of Li⁺ over K⁺ by clay minerals.

The strong negative correlations between Na⁺ and Mg²⁺, Ca²⁺, and Cl⁻ (r = −0.889/−0.732/−0.789) may have corresponded to the cation exchange of Na⁺ by Ca²⁺ and Mg²⁺ during the brine mixing process. When the underground brine was in a supersaturated state, the first to precipitate was NaCl. The negative correlation between SO₄²⁻ and Mg²⁺, Ca²⁺, and Cl⁻ (r = −0.689/−0.704/−0.671) may have reflected that the dissolution of calcium sulfate and magnesium sulfate was not the predominant origin of Ca²⁺ and Mg²⁺ in brine [53]. The negative correlation between Rb and K (r = −0.671) was consistent with their separation characteristics in the weathering process of mica minerals while the negative correlation between Cs and Ca²⁺ (r = −0.521) may have indicated their specific adsorption behavior in plagioclase weathering systems. These interactions collectively depicted the elemental cycling map of multiphase interfaces, providing key constraints for revealing regional hydrogeochemical processes.

3.5. Geospatial Clustering via SOM

The results of the SOM were visualized using the Unified Distance Matrix (U-matrix) and component plane (Figure 6A). Similar colors in the graph represented positive correlations between elements while color differences indicated negative correlations [54]. In this study, to assess the model’s reliability, the quantization error (QE) and topographic error (TE) values were calculated. After running the model, the TE and QE values were found to be 0.012 and 0.247, respectively, indicating acceptable topology retention. The SOM classified six categories. The first group of elements, with Na⁺, SO₄²⁻, and Ni, showed a gradually rising trend in content from up to low; in the contrary, the second group of elements, with Cl⁻, K⁺, and Mg²⁺, showed a gradually decreasing trend in content from up to low; the high-content areas of the third group of elements, with Li⁺, B₂O₃, HCO₃⁻, CO₃²⁻, and As, were exhibited on the right side of the feature plane while the low-content areas were mainly distributed in the upper-left corner, showing a clear spatial boundary; the fourth category included Cr, Cu, Zn, Hg, and Pb, which exhibited similar color gradients on the characteristic plane, with the maximum content distributed in the upper-left corner while the low-content areas were on the right side; the fifth category included Br, Cs, and Rb, and the maximum content was concentrated in the lower-right side of the plane; the last category included Ca²⁺ and Sr, with the maximum content distributed in the upper-right corner.

The spatial heterogeneity of sampling points, found using the K-means clustering method, was divided into four clusters (I~IV) (Figure 6B). The numbers in the figure represent the sample point numbers and were visualized through GIS (Figure 6C). Cluster III had the largest sample size of 5, followed by Cluster I (n = 4), Cluster IV (n = 4), and Cluster II (n = 2). Cluster I was marked in the upper-left corner of the SOM plane. From the perspective of element distribution, Cluster I exhibited high concentrations of Pb, Zn, Hg, and Cu on the characteristic planes. The Hg content in the brine in the research area was low, and at some sampling points, the Pb contents were also low, reflecting the low background characteristics of the region. However, the severe fluctuations in Zn and Cu concentrations suggested the presence of local human activities, indicating that the production of salt lakes had a certain impact on the quality of brine.

The indicators that affected Cluster II were Li, HCO₃⁻, B₂O₃, CO₃²⁻, and As, among which B₂O₃, CO₃²⁻, and As were the main indicators that determined clustering (Figure 6A). The Cluster II sampling points were located near the water replenishment zone of the Qarhan Salt Lake. Generally, the accumulation of Li and B₂O₃ is linked to the supply of Kunlun Mountain hot springs [46], and the concentrations of CO₃²⁻ and HCO₃⁻ gradually increased with the direction of water flow. Therefore, Cluster II of the sampling points were mainly influenced by the source of the water replenishment process.

Na⁺, SO₄²⁻, and Ni had higher contents in Cluster III, especially Na⁺, which was particularly prominent in Cluster III in the lower-left corner of the SOM plane. In correlation analysis, Na⁺ and SO₄²⁻ demonstrated a strong positive correlation, showing that the dissolution of sodium sulfate minerals may have been the main ion source. Cluster IV was situated in the lower-right corner of the plane, mainly including sampling points of the Dabuxun Salt Lake water. Rb, Cs, and Br all showed high contents on this plane. During the formation of salt lakes, rocks in surrounding areas such as the Kunlun Mountains are subjected to weathering and erosion, and potassium-containing minerals are decomposed, releasing Rb⁺ into the supply rivers, eventually flowing into the salt lake [14]. Sr mainly exists in carbonate minerals such as celestite and strontianite. During the formation of salt lakes, these minerals may be dissolved by groundwater or surface water, releasing strontium Sr²⁺ into the brine. Br is easily soluble in water and does not easily form stable precipitates. During the long-term evaporation process of salt lake brine, the concentration of Br in the brine is often high [47,48]. There are abundant brine resources underground in the Qaidam Basin, which may contain high concentrations of Br and then enter the Qarhan Salt Lake. Therefore, Rb, Sr, and Br mainly came from the weathering and leaching of minerals.

3.6. Source Apportionment Using the PMF Model

In this study, we imported the sample element content and its uncertainty data into the PMF 5.0 software. The results showed that when the number of factors was 4, $Q_{Robust}$/$Q_{True}$ tended to stabilize and converge to the minimum value. In addition, the R² values of all heavy metals in the PMF model were higher than 0.8 (ranging from 0.85 to 0.99), and most samples had residuals between −3 and 3, indicating that the selected number of factors for analyzing the sources of elements in brine was reasonable.

The PMF model analysis identified four principal sources for elements in the study area (Figure 7a,b). Factor 1 emerged as the dominant contributor, with 22% of the total variance, characterized by significant loadings from Ca²⁺, CO₃²⁻, HCO₃⁻, Li, Sr, Br, Rb, Cs, and As. Usually, Ca²⁺ and CO₃²⁻ primarily originated from the dissolution of minerals like calcite and dolomite. The dissolution of sulfate minerals such as celestite is the main source of Sr [14]. The contents of Br in the Dabuxun Salt Lake were supplied through fluvial inputs and concentrated via evaporative processes; those in the Qarhan Salt Lake were recharged by the upwelling Ca-rich spring water from the northward fracture belt. The elevated Br concentrations in deep brine reservoirs primarily originated from water–rock reactions and the release of Br from sediments and organic substances [47]. In this study, the high Br contents were mainly distributed in the Dabuxun Salt Lake. The glacier snow on the Kunlun Mountains and the water in the Golmud River and Nalenggele River had low Br levels; after evaporation and concentration, part of the Br was supplied to the Qarhan Salt Lake [48]. The primary factor that significantly impacted the levels of As concentrations was rock dissolution [55]. Therefore, F1 represented the source of the weathering and leaching of minerals.

Factor 2 accounted for 16% of variance, predominantly influenced by Ca²⁺, Pb, Ni, Zn, Cu, and B₂O₃. Pb, Cu, Zn, and Ni are often present in carbonate rock layers in the Qaidam Basin, and during long-term weathering and water–rock interactions, these minerals are dissolved, releasing Ca²⁺, Pb, Ni, Zn, Cu, and B₂O₃. The low Ca²⁺ levels could be linked to fluvial inputs into the Qarhan saline basin. The Golmud River serves as the dominant recharge water for the Qarhan Salt Lake, and previous isotope results indicated that the Golmud River had a high content of B [1]. Meanwhile, Pb and Zn are tracers of transportation activities, and the Qinghai–Tibet Railway and Qinghai–Tibet Highway are important traffic hubs on the Qinghai–Tibetan Plateau, passing through the Golmud River [56]. Therefore, transportation activities constitute one of the sources of Pb and Zn in the river. In addition, Golmud City is an industrial city, and Ni, Zn, and Pb generated from fuel combustion are major sources for the Golmud River [42,57]. Therefore, F2 represented the source of river water input.

The third factor contributed 37% and showed strong associations with K⁺, Mg²⁺, Cl⁻, Br, Hg, and SO₄²⁻. Sedimentary rocks and salt deposits rich in potassium, chlorine, and sulfate are widely distributed in the Qarhan Salt Lake area. These sediments release K⁺, Cl⁻, and SO₄²⁻ during long-term weathering and groundwater dissolution. Trace elements such as Br and Hg may also be precipitated from salt-bearing layers, especially when salt minerals such as sylvite and carnallite are dissolved; they release trace elements that coexist with the major elements. Therefore, F3 was analyzed as dissolution of salt-bearing layers. Factor 4 had the highest contribution rate to Cr and Ni, reaching 25%. The overall content of Cr and Ni was low, with locally high content. The Qarhan Salt Lake is an industrial area; its Cr and Ni are primarily derived from industrial emissions [42]. This factor strongly correlated with industrial emissions from smelting operations and coal combustion. Therefore, F4 was analyzed as encapsulating industry activities.

4. Conclusions

This study revealed that the brine in the Qarhan Salt Lake area exhibited extremely high salt characteristics, the mean value of TDS was 332.22 g/L, and the water type was MgCl₂-NaCl. The concentrations of heavy metals and dispersed elements showed significant differentiation, with a high enrichment of Li (69.66 mg/L), B₂O₃ (317.80 mg/L), Zn (5697.56 μg/L), and Cu (714.28 μg/L), while the partial enrichment of Pb and Cr (maximum values were 111.15 and 40.75 μg/L) required attention. The positive correlation between Na⁺ and SO₄²⁻ (r = 0.843) indicated the contribution of Na₂SO₄·10H₂O dissolution while the enrichment of Li and B₂O₃ (r = 0.736) reflected the hot spring recharge effect. The negative correlation between heavy metals and carbonates (r = −0.532/−0.636) revealed the fixed effect of carbonate precipitation in alkaline environments. Multivariate analysis identified that the elements in salt lake brine mainly came from the river water supply, weathering and leaching of minerals, and dissolution of salt-bearing layers and were locally influenced by human activities. The study emphasized the dual attributes of the Qarhan Salt Lake as a “high-value resource reservoir” and a “fragile salt lake ecology”.

Based on the research results of heavy metals and dispersed elements in the brine of the Qarhan Salt Lake, the following comprehensive policies and monitoring strategies were suggested: delimiting the Li and B₂O₃ enrichment areas as strategic resource protection areas; strict restrictions on high-pollution industrial activities in the surrounding areas; the optimization of the evaporation process design, controlling water–rock interaction; and building a multi-source monitoring network. It was recommended to establish a heavy metal monitoring system based on geochemical data in the potassium and lithium development, achieving resource efficiency extraction and environmental risk prevention and control.