Study Area

As the world’s largest antimony mine, the Xikuangshan mine is situated in central China, specifically 13 km north of Lengshuijiang City. It spans across a vast area of 26 km² and is located within mountainous terrain that runs from northeast to southwest, with elevations ranging between 220 and 823.2 m. The climate is known for its high humidity, with an average annual precipitation of 1381.60 mm, evaporation reaching 903.30 mm, and temperatures averaging 16.7ºC from 1949 to 2012. The primary watercourses in the mining region that flow into the Zijiang River include the Xuanshan River, Qingfeng River, Feishuiyan Stream, Tanjia Stream, and Batangshan Stream. The Xikuangshan mine is internationally recognized as “The World’s Antimony Capital”, earning an impressive reserve of around 2.50 million tons of Sb [1, 17, 47, 48].

The Xikuangshan Sb mine area is situated in a hydrogeologically isolated region, enclosed by the NE30°-oriented fault NO.75 from the east and west. It encompasses a lamprophyre vein spanning approximately 10 km. As shown in Fig. 1, the primary aquifers in the mining region consist of the lower Shetianqiao karst aquifer (referred to as D₃s² water) and the upper Magunao karst aquifer (known as D₃x⁴ water), which have been extensively studied by Wen et al. [5] (2016) and Hao et al. [1, 6]. The Sb ore bodies are located at the top of the silicified limestone layers of the D₃s² water, which have no hydraulic connection with D₃x⁴ water. The D₃x⁴ water, which consists of limestone and sandy limestone with an average thickness measuring 258 m and a hydraulic conductivity of 0.0092 m/d as reported by Wen et al.’s study conducted in 2016 and updated in 2022, plays a crucial role as a primary drinking water source for local residents. The main recharge source for the D₃x⁴ water is infiltrated precipitation and agricultural irrigation, while groundwater flows from southeast to northwest in the North Mine and from northeast to southwest in the South Mine, respectively. D₃x⁴ water discharge occurs through springs that crop out substantially and drainages associated with mining activities [4]. The primary hydrochemical facies in unpolluted D₃x⁴ water is Ca-HCO₃-SO₄ type, and Sb concentration is less than 15 mg/L [4]. Some dwellings, industrial facilities, waste rock, and slag are situated over the D₃x⁴ water [6, 49]. Moreover, mining regions extensively cultivate rice, corn, and vegetables as their main agricultural produce [17, 49].

Sample Collection

Based on the previous studies conducted on hydrogeology and geochemistry, a total of 20 springs were selected for sampling D₃x⁴ water in February 2023 (Fig. 1a). Prior to collection, the brown plastic sampling bottles underwent sequential rinsing with distilled water and D₃x⁴ water samples. All water samples were collected in the field and ultrafiltered sequentially through MilliPore filters (high-density polyethylene (HDPE)) with sequential pore sizes of 0.45-μm, 100- kDa, and 1-kDa by a cross-flow ultrafiltration system (CFUS, Sartorius Vivaflow®.200) [50].

Fig. 1.

Location of the study area and distribution of the D₃x⁴ water sampling sites from the Xikuangshan mine.

Each of the filtered samples was separated into three categories: high-molecular-weight DOM (HDOM) D₃x⁴ water (fraction<0.45 μm), middle-molecular-weight DOM (MDOM) D₃x⁴ water (fraction<100 kDa), and low-molecular-weight DOM (LDOM) (fraction<1 kDa). A total of 60 D₃x⁴ water samples were obtained during the process of ultrafiltration. The experiment involved utilizing a polymer-enhanced ultrafiltration system with a capacity of 300 mL by pressurizing and filtering with N₂ gas. The ultrafiltration process was previously documented by Dundar et al. [51] and Li et al. [33]. To facilitate the analysis, the collected samples for the major cation and Sb analyses were treated with ultrapure HNO₃ at a ratio of 1:1 (v/v) until reached a pH<2.0, after which they were promptly stored at a temperature of 4ºC in darkness. To prepare samples for stable organic carbon isotope analysis (δ¹³C_DOC), they were acidified using 85% H₃PO₄ until the pH was less than 2. For stable inorganic carbon isotope analysis(δ¹³C_DIC) was sterilized with HgCl₂ and then collected in HDPE brown glass bottles with 100 mL.

Sample Analysis

The pH and total dissolved solids (TDS) were measured on-site at a sampling location using a portable pH meter (HANNA H18424, Italy) and a portable conductivity meter (HANNA H1833, Italy), respectively. The concentrations of bicarbonate ions (HCO₃^-) were determined by acid-base titration during field measurements with an analytical precision of 1.0 mg/L. Sb concentrations were determined using a hydride generation atomic fluorescence spectrometer (Qingdao) with a relative standard deviation of ±5% and an analytical precision of 0.10 µg/L [17, 28]. Total Fe (TFe) levels were determined using spectrophotometry (DR2800, HACH, USA) and the phenanthroline method with an analytical precision of 0.01 µg/L [17, 28, 52]. δ¹³C_DOC and δ¹³C_DIC in all samples were determined using isotope ratio mass spectrometry (Trace GC Ultra, Thermo Fisher Scientific, USA) coupled with an online high-precision gas headspace sampler, GasBench (Thermo Fisher Scientific), following the methods of Yu et al. [53] and Zhou et al. [54] with precisions<0.2‰.

The level of DOM was assessed utilizing a TOC- 5000 total organic carbon analyzer from Japan and represented as dissolved organic carbon (DOC), with an analytical accuracy of 0.01 mg/L. The UV-visible spectrophotometer (Hach DR-5000, USA) and the three-dimensional fluorescence spectrophotometer (F7000, Japan) were utilized to perform measurements of ultraviolet-visible and fluorescence in a 10-mm quartz cuvette. Both devices were equipped with a 150-W xenon lamp that does not emit ozone and operated at a consistent temperature of 20ºC. The emission wavelength (EM) was scanned from 220 to 550 nm while the excitation wavelength (EX) ranged from 200 to 400 nm, with sampling intervals of 5 nm during the excitation emission matrices (EEMs) analysis. The fluorescence measurements were conducted with an excitation wavelength step size of 5 nm and an emission wavelength step size of 2 nm. The scan rate was set at 2400 nm/min, the slit width was adjusted to 5 nm, and the amplification voltage used was 700 V.

The spectral overlap was determined by employing PARAFAC modeling with the assistance of fluorescence components from the DOMFluor database [55, 56]. All mean fluorescence intensities are calculated based on three repeated analyses, with subsequent subtraction of blank values and final normalization to Raman units (R.U.). The split half analysis and residual analysis were used to verify the reliability of the three-component model [33]. Residual intensities, were obtained by subtracting the PARAFAC-modeled EEM from the measured EEM maximum deviation of 10% compared to the measured EEM intensities across all samples.

To distinguish DOM sources, the fluorescence index (FI) was determined by dividing the measurement at an emission wavelength of 450 nm by that at 500 nm, after being excited at 370 nm, providing a metric for distinguishing DOM derived from terrestrial and microbial sources [32, 43, 57]. Furthermore, the biological index (BIX), which serves as a measure of indigenous biological activity in aquatics, was determined by dividing the emission intensities at wavelengths of 380 nm and 430 nm under a consistent excitation at 310 nm [58, 59]. In addition, the ratio of the emission scan at 435-480 nm to the emission scan at 300–345 nm with an excitation at 254 nm was utilized for evaluating the humification index (HIX), which serves as an indicator of DOM humification [60-62]. The calculation of the specific ultraviolet absorbance (SUVA₂₅₄), which is linked to the aromatic nature of organic substances, involved normalizing the UV absorbance at 254 nm with respect to the DOC concentration [36, 59].

To enhance further characterization, the DOM was segregated into two categories. Initially, protein:humic ratios were computed by determining the proportion of protein-like DOM constituents in relation to the total sum of humic-like DOM components. Secondly, the microbial:terrestrial ratios were determined by calculating the proportion of DOM components derived from microorganisms compared to that of terrestrially derived DOM components.

Statistical Analysis

The statistical analysis software, Origin 2021, was employed to establish correlations among all findings. The PARAFAC analysis was performed using the DOM Fluor v.1.7 toolbox in MATLAB (Natick, MA, USA). Descriptive data can be found in Table 1, while Table 2 provides information on the spectroscopic characteristics of DOM components.

General Hydrochemistry Characterization

of D₃x⁴ Waters with Different Molecular Weights

According to the geogenic values without pollution [5, 6], the D₃x⁴ water samples were categorized into two groups: low-Sb groundwater (<15.00 µg/L) and high-Sb groundwater (>15.00 µg/L). pH values ranging from 7.14 to 9.74 with a mean of 8.03 were observed (Table 1), indicating a weakly alkaline environment in all D₃x⁴ waters. The phenomenon indicated that H⁺ produced from stibnite oxidation had been immediately neutralized by carbonate minerals in D₃x⁴ waters. The TDS concentrations in the high-Sb D₃x⁴ water ranged from 121 to 1515 mg/L, with an average of 542 mg/L, surpassing those found in the low-Sb D₃x⁴ water samples (ranging from 254 to 331 mg/L, with an average of 300 mg/L).

The concentration of Sb showed no significant differences among HDOM, MDOM, and LDOM in the low-Sb D₃x⁴ water (Fig. 2a). There was a slight decrease during sequential ultrafiltration in the high-Sb D3x4 water, and a slight decrease was observed during sequential ultrafiltration in the high-Sb D₃x⁴ water, suggesting that Sb predominantly existed as a truly dissolved form in the LDOM D₃x⁴ water. These findings align with the research conducted by Zhang et al. [63] and Jia et al. [50] on Sb migration as well as the study by Li et al. [33] on As presence in shallow groundwater.

A decrease in DOM concentration was noted in LDOM and MDOM D₃x⁴ waters compared to HDOM D₃x⁴ water (Fig. 2c), suggesting that the HDOM D₃x⁴ water exhibited the highest level of DOM content. The majority of TFe detected in the extracts was observed in the HDOM D₃x⁴ water (Fig. 2b), which experienced a significant decrease in both MDOM and LDOM D₃x⁴ waters. This suggested that TFe predominantly existed as large colloids within the HDOM D₃x⁴ water [33, 35, 50, 63, 64]. The consistent trends between DOM and TFe indicate that TFe may have a higher affinity for binding with HDOM in the D₃x⁴ water [35]. Earlier research had established that substances found in HDOM, such as humic-like and fulvic acids, exhibited a preference for forming DOM-Fe complexes with Fe hydroxides in groundwater [17, 35, 65, 66]. Therefore, it can be deduced that TFe predominantly exists as colloidal particles attached to HDOM in the D₃x⁴ water [40]. However, the distribution of Sb in the D₃x⁴ water with varying pore sizes exhibited distinct variations compared to the distributions of TFe and DOC as depicted in Fig. 2(a-c), indicating that HDOM was not a crucial factor for promoting Sb enrichment in the D₃x⁴ water.

Fluorescence Characteristics of DOM with Different Molecular Weights

Fig. 3 showed the six main EEM peaks: Ex 266 (336)/Em 435 (Peak A, representing the C1 component) for the terrestrial humic-like component [37, 67, 68], Ex 278/Em 336 (Peak T, reflecting the C2 component) for the tryptophan-like substance [39, 69], Ex 306/Em 385 (Peak M, C3 component) for the microbial humic-like component [70, 71], Ex 266/Em 296 (Peak B, C4 component) for the carboxylic and phenolic groups component [32, 72], Ex 266/Em 507 (Peak C, C5 component) for the marine humic acid-like component [73, 74], and Ex 400/Em 464 (Peak D, C6 component) for the fulvic acid component [34, 64].

Fig. 2.

Box-whisker plots of Sb a), TFe b), and DOC c) for different size fractions of DOM in D₃x⁴ water.

Fig. 3.

Spectral properties of the six fluorophores identified by the PARAFAC analysis.

SUVA₂₅₄, BIX, FI, and HIX of DOM with different molecular weights were shown in Fig. 4. During the process of sequential ultrafiltration, there was a slight increase observed in BIX, SUVA₂₅₄, and HIX for low-Sb D₃x⁴ waters (Fig. 4a–d), while a significant decrease was observed for high- Sb D₃x⁴ waters. Significantly, the HDOM exhibited mean values of SUVA₂₅₄ and HIX that were 2.65 and 1.27 times higher in the HDOM D₃x⁴ water than those observed in the LDOM D₃x⁴ water, respectively, suggesting a more pronounced presence of macromolecular aromatic substances in the HDOM D₃x⁴ water. The rate of increase in BIX was found to be higher in the low-Sb D₃x⁴ water compared to the high-Sb D₃x⁴ water during the sequential ultrafiltration process (Fig. 4b), indicating the presence of biological and aquatic bacterial sources of DOM played a significant role in enriching Sb levels in the low-Sb D₃x⁴ water [17, 27]. The FI values of both low- and high-Sb D₃x⁴ water samples exhibited a wide range from 1.40 to 1.90, indicating the presence of terrestrial and microbial contributions (Fig. 4c). This suggests that the main sources of DOM in the D₃x⁴ water are likely dominantly derived from terrestrial and microorganism origins [37, 43]. During the sequential ultrafiltration process, there was a slight transition in the FI values from terrestrially and microbially derived zones to microbially derived zones in high-Sb D₃x⁴ water (Fig. 4c), suggesting a greater contribution from microbial sources in LDOM D₃x⁴ water.

As indicated in Table 2 and Fig. 5a), the PARAFAC components of DOM were dominated by humic-like and fulvic acid components (the combined percentages of C1, C3, C5, and C6), followed by protein-like components (the sum percentages of C2 and C4) in HDOM and MDOM D₃x⁴ waters, whereas the contents of protein-like components considerably increased and became predominant in the LDOM D₃x⁴ water. The observation provided additional evidence that the protein-like constituents exhibited a progressive increase in dominance and promotion in the LDOM D₃x⁴ water, whereas humic-like and fulvic acid components were predominantly present in HDOM D₃x⁴ water throughout the sequential ultrafiltration process [17, 75]. Previous studies have established that the majority of humic-like and fulvic acid components are present in the colloidal fraction, which can be effectively captured through ultrafiltration utilizing membranes with a molecular weight cut-off of 1 kDa [35].

In Fig. 5b), the percentages of C2 and C4 significantly increased with the decrease in ultrafilter pore size, while the abundances of C1, C3, C5, and C6 declined during the sequential ultrafiltration process. For the given samples, there was a significant increase in the proportion of C4, accompanied by notable decreases in abundances of C1 and C3, when compared to the relative proportions of other components in the LDOM D3x4 water. This indicated that the LDOM D₃x⁴ water contained a higher proportion of the tyrosine-like component, while the HDOM and MDOM D₃x⁴ water favored the presence of humic-like components as large colloids [3, 40, 64]. The observed variations in the relative of the humic-like and fulvic acids, as well as protein-like components, were attributed to changes occurring during the sequential ultrafiltration process in Fig. 5a) and Table 2. As depicted in Figure 5c, the proportions of protein:humic and microbial:terrestrial progressively rose with the reduction in ultrafilter pore size, providing confirmation that the LDOM D₃x⁴ water primarily contained a protein-like component derived from microorganisms.

In general, the HDOM D₃x⁴ water exhibited higher levels of HIX, SUVA₂₅₄, and percentages of C1 and C3, while lower levels of BIX, FI, and percentages of C2 and C4 were observed compared to the LDOM D₃x⁴ water. This suggests that HDOM exhibited elevated levels of humification and aromaticity levels whereas LDOM was influenced by the microbial origin of DOM [33]. A similar outcome was observed in the high-As shallow groundwater with varying molecular weights in the Hetao Basin [33, 76], demonstrating that the chemical characteristics of DOM with distinct molecular weights may have similar effects on the fates of As and Sb.

Fig. 4.

Box-whisker plots of SUVA₂₅₄ a), BIX b), FI c), and HIX d) for different size fractions of DOM in D₃x⁴ water.

DOMs with Different Molecular Weights as a Complexing Agent for Sb

As depicted in Fig. 2a) and b), the majority of TFe was observed as Fe hydroxides in the HDOM D₃x⁴ water, while Sb was predominantly present in the LDOM D₃x⁴ water. Meanwhile, there were slight correlations between Sb, DOC, and TFe in the HDOM D₃x⁴ water (Fig. 6a) and a significant positive correlation (R² = 0.48) was found between Sb and DOC in the LDOM D₃x⁴ water (Fig.6b), suggesting that the presence of enriched TFe hydroxides had minimal impact on the migration of Sb within the LDOM and MDOM D₃x⁴ water. This contrasted with the findings of Li et al. [33] in their study on a high As groundwater from the Hetao basin, where they observed that higher binding capacity for As was demonstrated by larger Fe colloids (>10 kDa) during the sequential ultrafiltration process. The phenomenon implied that As exhibited a tendency to form stronger associations with Fe complexes compared to Sb, owing to DOM with different molecular properties.

Compared to low-Sb D₃x⁴ water samples, TFe mean concentrations in high-Sb D₃x⁴ water samples were found 3.14 times higher, indicating that Sb was likely associated with TFe hydroxides and complexes formed by DOM in high-Sb D₃x⁴ water samples. Humic-like components could form DOM-Fe-Sb complexes by complexing with Fe hydroxides and Sb [17, 23, 63, 77]. This finding aligns with the observed higher values of HIX and SUVA₂₅₄, as well as lower values of BIX and protein:humic ratio in the HDOM D₃x⁴ water sample (Fig. 4a, b, d, and Fig. 5a). Complexes between Sb and humic-like substances of DOM are typically formed via the presence of positively charged amino groups in the DOM or metal cation bridges [20, 26, 40]. The interaction strength and durability of the binding between DOM and Sb were notably increased in groundwater through ligand substitution, the creation of negatively charged complexes, and the presence of bound hydrogen bridges [63, 78, 79].

Fig. 5.

Box-whisker plots of CX percentages for different size fractions of DOM in D₃x⁴ water.

Fig. 6.

a) Sb versus TFe concentration in the HDOM D₃x⁴ water and b) Sb versus DOC concentration in the LDOM D₃x⁴ water.

In addition, the Sb concentration in the LDOM D₃x⁴ water showed weak (R² = 0.14) and moderate (R² = 0.32) positive correlations with the percentages of C2 and C4, respectively in Fig.7a and b. This suggested that protein-like component substances of DOM with lower HIX and SUVA₂₅₄ and higher percentages of C2 and C4 may contribute to an increase in Sb levels in the LDOM D₃x⁴ water. These protein-like substances are known for their strong affinity, such as copper and cadmium [20, 36, 80, 81]. Furthermore, the stable Sb(III) could be rapidly oxidized to an easily mobile Sb(V) in the presence of the quinone groups in neutral and alkaline waters, resulting in Sb concentration enrichment in the LDOM D₃x⁴ water [23, 78, 82].

The changes in fluorescent intensity for the six PARAFAC-derived components during the increase in Sb concentration were illustrated in Fig. 7c), specifically focusing on the LDOM D₃x⁴ water. The quenching curves exhibited a significant correlation with the origins of DOM sources. The initial decrease in fluorescent intensities of C4 and C2 was followed by a gradual decrease and stabilization as the concentration of Sb increased. Conversely, there was a slight increase in the fluorescent intensities of C3 and an increase in Sb concentration, which could be attributed to the combination of Sb with DOM [19, 83]. Weak quenching effects were observed for C1 and C6. These results suggested that the protein-like substances had more significant quenching effects than the humic-like materials in the LDOM D₃x⁴ water.

In addition, the quenching levels of C4 were pronounced compared to those of C2, suggesting that the carboxylic and phenolic groups exhibited a stronger affinity for Sb in the LDOM D₃x⁴ water [58, 81, 83]. Generally, the abundant presence of carboxylic and phenolic groups in DOM led to its high binding affinity towards heavy metals [58, 84], indicating the C4 component in the LDOM D₃x⁴ water may be attributed to higher carboxylic and phenolic groups than other components. Moreover, the presence of carboxylic and phenolic groups on the C4 component may potentially hinder the adsorption onto TFe hydroxide surfaces, leading to the liberation of previously adsorbed Sb into the D₃x⁴ water [17, 85].

Bioreactivity of LDOM and Microorganism Roles

Previous studies had proved the substances resembling proteins, characterized by a high BIX and low HIX and SUVA₂₅₄ values, along with significant bioreactivity, were found to enhance the Sb mobility in groundwater systems by facilitating electron transfer and energy acquisition for microbial degradation and activities [30]. Generally, microbial degradation of organic matter causes a lower δ¹³C_DIC and higher δ¹³C_DOC in groundwater, reflecting an active microbial process [54, 86, 87].

As depicted in Fig. 8a), the <1 kDa fractions exhibited elevated δ¹³C_DOC values and almost unchanged δ¹³C_DIC values during the sequential ultrafiltration process, suggesting a more pronounced microbial influence on DOM within the LDOM D₃x⁴ water. The observed fluctuations in δ¹³C_DOC may be ascribed to the surface-driven microbial oxidation of organic substances occurring within the shallow groundwater. Extensive regions of rice, corn, and vegetables were allocated within the upper weathering zone of the D₃x⁴ water. The input of DOM from precipitation may stimulate microbial respiration, resulting in an increase in δ¹³C_DOC levels [88]. On the other hand, Sb with different valence states served as substrates for electron transfer and energy acquisition by autotrophic species in the oligotrophic groundwater environment [2, 30, 89]. The TFe levels in the LDOM D₃x⁴ water varied from undetectable to 30.00 µg/L (with an average of 2.59 µg/L), creating favorable conditions for microbial degradation through its interaction with Fe oxides and DOM [33, 89]. Furthermore, oxidative stibnite, serving as the primary mineral containing antimony, could have potentially acted as a significant energy provider for autotrophic communities through the discharge of antimony and sulfur compounds into the D₃x⁴ water environment [2, 30].

Fig. 7.

a) Percentage of C2, b) percentage of C2, and c) fluorescence intensity versus Sb concentration in the LDOM D₃x⁴ water.

In Fig. 8b), a noticeable decline in the difference between δ¹³C_DIC and δ¹³C_DOC was observed in the LDOM D₃x⁴ water as the Sb content increased (R² = 0.72), suggesting Sb mobilization was related to the degree of microbial degradation of DOM [54, 90]. The lower difference between δ¹³C_DIC and δ¹³C_DOC suggests a higher level of microbial degradation of DOM, even though the isotopic variation in organic carbon is minimal and can be disregarded during microbial degradation processes [54, 91, 92]. The correlation between δ¹³C_DIC and the difference between δ¹³C_DIC and δ¹³C_DOC was found to be significantly positive, indicating that the oxidative decomposition of organic carbon played an important role in the reduction in δ¹³C_DIC. In the process of microbial degradation of DOM, microorganisms exhibited a preference for lighter carbon isotopes (¹²C), leading to decreased δ¹³C_DIC levels and increased concentrations of HCO3- in groundwater [80, 93]. Compared to the low-Sb D₃x⁴ water, the high-Sb D₃x⁴ water exhibited lower and more negative δ¹³C_DIC subtracted by δ¹³C_DOC, indicating that the microbial degradation of DOM played a crucial role in the mobilization and enrichment of Sb in the LDOM D₃x⁴ water [2, 17, 30].

Fig. 8.

a) Box-whisker plots of CX percentages for different size fractions of DOM, b) relationship between δ¹³C_DIC - δ¹³C_DOC and Sb concentration, and c) relationship between δ¹³C_DIC - δ¹³C_DOC and δ¹³C_DIC in the LDOM D₃x⁴ water.

Fig. 9.

The sequential ultrafiltration process of DOM with different molecular weights for Sb migration in D₃x⁴ water.

Therefore, our results suggested that Sb was primarily bound to protein-like substances with biological activity and is influenced by microbial degradation in the LDOM D₃x⁴ water. As a result, the changes in Sb concentration were not found to be significant throughout the sequential ultrafiltration process (Fig.9). However, the impact of TFe was not ignored, which could play a role in facilitating the combination of Sb and DOM. This combination forms a Fe bridge, thereby promoting the accumulation of Sb in HDOM and MDOM D₃x⁴ waters.