Hydrogeochemical Facies and Health Hazards of Fluoride and Nitrate in Groundwater of a Lithium Ore Deposit Basin

Abstract

Fluoride and nitrate contamination in groundwater is a global concern due to their toxicity and associated negative health effects. This study incorporated a comprehensive methodology, including hydrogeochemical analysis, drinking and irrigation water quality assessment, source apportionment, and health risk estimation of groundwater fluoride and nitrate in a lithium ore deposit basin in western Serbia. Groundwater major ion hydrogeochemistry was governed by water–rock interactions, with Ca-Mg-HCO₃ identified as the predominant groundwater type. The entropy-weighted water quality index (EWQI), sodium adsorption ratio (SAR), and sodium percentage (%Na) revealed that 95% of the samples were of excellent to good quality for both drinking and irrigation. Moreover, the results showed that fluorides were of geogenic origin, whereas nitrates originated from agricultural activities. Although the fluoride and nitrate levels in groundwater were relatively low, averaging 1.0 mg/L and 11.1 mg/L, respectively, the results of the health risk assessment revealed that the ingestion of such groundwater can still lead to non-cancerous diseases. The threshold of one for the hazard index was exceeded in 15% and 35% of the samples for adults and children, respectively. Children were more vulnerable to non-carcinogenic risk, with fluorides being the primary contributing factor. The study outcomes can serve as a reference for other lithium-bearing ore areas and guide the management of regional groundwater resources.

1. Introduction

Groundwater serves as the primary source of both drinking and irrigation water, and its quality is closely associated with human health, ecosystem stability, food safety, and socioeconomic progress [1,2,3]. However, population expansion, urbanization, industrialization, and agricultural and mining development have collectively contributed to global groundwater deterioration [4,5,6,7]. Moreover, the demand for groundwater has been growing consistently over the last several decades, with about two-thirds of people worldwide currently being affected by water scarcity [8].

Nitrate and fluoride are among the most prevalent contaminants found in groundwater [9], with concentrations exhibiting significant global variability. For instance, between 1992 and 2012, the average nitrate level reported in many European countries was 17.5 mg/L [10]. Nevertheless, depending on specific local factors, concentrations can be several orders of magnitude greater, such as those found in Saudi Arabia that reached 108 mg/L [11] or in China that reached 329 mg/L [2]. On the other hand, according to Jha and Tripathi [12], fluoride concentrations in groundwater can naturally range from trace amounts to exceed 25 mg/L.

The common occurrence of nitrate in groundwater is driven by its high solubility and difficulties with its soil fixation [13]. Nitrates can be found in groundwater as a result of atmospheric deposition or the symbiosis of certain plants; however, they predominantly arise from anthropogenic activities [14]. More precisely, agricultural activities and excessive use of nitrogen fertilizers stand out as the main sources of nitrates [15]. Additionally, nitrate groundwater contamination can arise from improper industrial wastewater disposal and the unregulated discharge of domestic sewage and animal waste [16,17,18]. High nitrate levels in groundwater can contribute to eutrophication when these nutrients enter water bodies [19]. This process stimulates excessive algal and aquatic plant growth, leading to oxygen depletion, harmful algal blooms, and negative impacts on aquatic ecosystems [15]. Apart from the negative effects on the environment, prolonged use of water containing elevated nitrate concentrations can lead to adverse health effects, including methemoglobinemia, thyroid dysfunction, and neonatal abnormalities [4,20,21]. For the above reasons, the World Health Organization (WHO) has prescribed a maximum permissible nitrate concentration in drinking water of 50 mg/L [22].

Conversely, fluorides are essential for maintaining human health [23,24]. At certain permitted levels, they have a beneficial effect on dental health and can prevent caries [25]. However, excessive levels can cause numerous health issues, primarily dental and skeletal fluorosis [26]. People are predominantly exposed to fluorides through drinking water, accounting for more than 75% of fluoride intake [1]. To mitigate this exposure, the WHO [22] set the maximum permissible level of fluorides in drinking water to 1.5 mg/L. However, around 200 million people in 100 countries are affected by fluoride pollution in groundwater [27]. Fluorides in groundwater are predominantly of geogenic origin, coming from the weathering of fluorine-rich minerals, such as apatite, hornblende, and biotite [28]. In addition to natural sources, anthropogenic activities such as coal burning, ceramic and glass production, and aluminum processing also contribute to groundwater pollution by releasing fluorides [29,30]. Therefore, ensuring a safe water supply requires identifying the distribution and sources of contaminants, assessing groundwater quality, and evaluating health risks. Such investigations are vital for public health and provide crucial information for decision-makers when implementing measures to mitigate groundwater degradation.

To investigate groundwater pollution, its quality, and health risks, scientists rely on various chemometric techniques, such as geostatistical analyses [31,32], principal component analysis [33,34], positive matrix factorization [35,36], and Monte Carlo simulation [37,38]. Multiple approaches are often used simultaneously in a single research project due to the high degree of complementarity or competition among them [39]. Drinking water quality is commonly evaluated using the water quality index (WQI) [40,41,42]. However, this method can introduce significant subjectivity, resulting in less accurate outcomes. The entropy-weighted water quality index (EWQI) resolves this issue by providing a more objective and accurate assessment by weighting each index according to its entropy value [43]. Considering that water is capable of carrying up to 70% of pollutants and is linked to approximately 20% of cancer cases [22], it is extremely important to assess the health risks that groundwater pollutants pose to humans. For this purpose, scientists have widely used the method proposed by the United States Environmental Protection Agency (USEPA) [44], which is based on a deterministic approach. Taking into account the chronic daily intake of each contaminant, this method can quickly and easily quantify non-carcinogenic and carcinogenic risks in various population groups [11,45,46]. Yet, there are certain drawbacks to this approach, such as the high degree of uncertainty regarding its parameters, which may cause the risks to be overstated or understated [47]. As a probabilistic approach, Monte Carlo simulation can resolve this problem by using input parameter distributions to reduce uncertainties [48]. In this way, health risks are presented in the form of probabilistic distributions, thereby providing more accurate results [49].

This study incorporated various methods to investigate groundwater hydrogeochemistry, drinking and irrigation quality, pollution sources, and health risks arising from nitrate and fluoride groundwater drinking in a lithium ore deposit area. Lithium research is already being conducted in this area, and there are plans for possible mine development. Lithium is a critical element due to its essential role in producing rechargeable batteries, which power a wide range of devices, from smartphones and laptops to electric vehicles, and renewable energy storage systems. The increasing demand for sustainable energy solutions and the transition to electric transportation have significantly boosted the global demand for lithium, making it a strategic resource for the future [50,51,52]. More specifically, the demand for lithium rose from 37,000 tons in 2016 to 52,000 tons in 2018, with forecasts indicating it will exceed 380,000 tons by 2028 [53].

As such, it is crucial to thoroughly investigate the environmental conditions at sites where lithium mines are planned to open. This ensures that mining activities do not adversely affect local ecosystems, water quality, and human health and helps in developing strategies to mitigate potential environmental impacts. Understanding the baseline environmental state will allow for more responsible and sustainable lithium extraction practices, safeguarding both the environment and the communities surrounding these mining operations. To the best of our knowledge, this type of research has not been conducted in such an area. This research seeks to improve comprehension of the hydrochemistry and water quality of a lithium-rich ore area, where groundwater serves as the main source of drinking water, focusing on the nitrate and fluoride contents in groundwater and the related health risks. The findings of this study can serve as a reference for similar areas elsewhere and can offer guidance for managing regional water resources and ensuring drinking water safety.

2. Materials and Methods

2.1. Study Area

The study area (Figure 1) is situated in western Serbia, in southeastern Europe. It belongs to the Kolubara district, named after the Kolubara River, whose catchment comprises 3638.5 km². The Kolubara River originates in the city of Valjevo, flowing from west to east over a length of 86.4 km before merging with the Sava River near the city of Obrenovac. The study area is characterized by a lowland to hilly topography, with the altitude increasing with distance from the Kolubara river and an average elevation of approximately 272 m [54]. The Kolubara river, along with its numerous tributaries, flowing through the area, created fertile alluvial soil suitable for cultivation. As a result, the majority of the population is engaged in agriculture. Moreover, agricultural land accounts for more than 50% of the total area, primarily fields and arable land [55].

The study area is a part of the Valjevo-Mionica lithium-ore deposit basin. The lithium mineralization in Valjevo, comprising borax, probertite, and lithium clays, was found within marine carbonate-clastic facies at depths of 240 to 315 m beneath ground level. This mineralization comprises two distinct zones: the higher stratigraphic lithium zone with Li₂O concentrations averaging 0.17 wt%, and the main zone, which extends approximately 32 km² and has average concentrations of Li₂O of 0.16 wt%. Initial estimates of reserves indicate a total of 10 million tons of Li₂CO₃ [56].

The study area belongs to the Vardar zone tectonic unit, characterized by its complex geological history and diverse rock formations. The region features crystalline schists, Carboniferous sedimentary deposits, and Jurassic igneous rocks. It also includes Triassic hemipelagic and eupelagic sediments and diabase–chert rocks. The flysch of the Cretaceous age and calc-alkaline volcanic rocks of the Tertiary age further contribute to its complexity [57].

Hydrogeologically, the study area is distinguished by a surface layer of clayey humus, underlain by dusty clay, and subsequently layers of medium- to coarse-grained gravel, fine-grained sand, coarse-grained clayey gravel, and marly clay [58]. Accumulation of groundwater occurs within sandy gravel deposits located in Kolubara river channels. The estimated volume of groundwater held within its alluvial formations accounts for 130 L/s. Generally, the Kolubara River alluvium demonstrates variability in thickness, typically being relatively small. Groundwater recharge and discharge are naturally managed through the Kolubara River and its tributaries, establishing a hydraulic connection between the aquifer and the rivers. Moreover, groundwater discharge is facilitated by water extraction wells, while precipitation contributes to groundwater recharge [59].

2.2. Sampling and Analysis

In the study area, 20 groundwater samples (Figure 1) were collected in 2020 from wells and springs and analyzed for 19 physico-chemical parameters (pH, temperature (T), electrical conductivity (EC), total dissolved solids (TDS), dissolved oxygen (DO), HCO₃⁻, F⁻, Cl⁻, SO₄²⁻, NO₃⁻, B, SiO₂, Sr²⁺, Ca²⁺, K⁺, Li⁺, Mg²⁺, Na⁺, and PO₄³⁻). These wells and springs are used by local rural residents for domestic and drinking purposes, with the exception of sample no. 5, a well that was dug to prove lithium deposits.

Water sampling procedures utilized 1 L polyethylene bottles. To minimize disruption to well water and ensure accurate sampling, a low-flow submersible pump (Geosub 2, Geotech Environmental Equipment, Inc., Denver, CO, USA) was utilized at a flow rate of 80 mL/min. Prior to sampling, groundwater at each well sampling site was pumped out for 15 min to reduce stagnant water interference. Bottles were cleaned sequentially with deionized water and sample water, followed by filtration using 0.45 µm Millipore filters. Groundwater parameters, including pH, T, EC, TDS, and DO, were determined at sampling sites using a portable multi-probe meter (YSI 650 MDS, YSI Inc., Yellow Springs, OH, USA). Following collection, samples were stored at 4 °C in the dark before laboratory examination. The analysis of cations (Na⁺, K⁺, Ca²⁺, Mg²⁺, Sr²⁺, Li⁺) and anions (NO₃⁻, Cl⁻, SO₄²⁻) was conducted using an ion chromatography system (Dionex DX-500, Thermo Scientific, Waltham, MA, USA). Acid–base titration utilizing 0.1 M HCl was used for HCO₃⁻ determination. A fluoride ion-selective electrode (YSI TruLine, YSI Inc., Yellow Springs, OH, USA) was used to determine F⁻, while PO₄³⁻ was determined by VIS spectrophotometry (DW-721G, Drawell, Chongqing, China). Inductively coupled plasma optical emission spectrometry (ICP-OES iCap 6500 Duo, Thermo Scientific, Waltham, MA, USA) was used to determine the boron and silicon contents in groundwater samples.

Strict quality assurance and control protocols were implemented. All analyses were performed in accordance with the protocols outlined in APHA [60] and performed in triplicate. Concentrations are reported as mean values between triplicates, with the relative standard deviation below ±5%. During the laboratory analyses, reagent blanks were utilized after every five samples, and spiked water samples were included for recovery studies, with the recovery rates ranging from 92% to 106%. Moreover, to assess the accuracy of the physico-chemical analyses, the ion charge balance error (%CBE) comprising the total cation concentration (TC, meq/L) and total anion concentration (TA, meq/L) (Equation (1)) was employed. The %CBE values obtained were within the acceptable range of ±5% for each sample.

$$\%CBE={\displaystyle \frac{TC-TA}{TC+TA}}\times 100\%$$

(1)

2.3. Entropy-Weighted Water Quality Index

For assessing the quality of groundwater for drinking, the entropy-weighted water quality index (EWQI) methodology was utilized. Calculation of EWQI involves five steps, as follows [21,61]. The initial step in this method involves constructing the eigenvalue matrix X (Equation (2)), based on a set of m groundwater samples and n chemical parameters. Afterwards, using Equation (3) for standardization, where (x_ij)_max and (x_ij)_min denote the maximum and minimum values of chemical parameter j in sample i in matrix X, the initial matrix X is transformed into the standardized matrix Y (Equation (4)). In the third step, the entropy (e_j) and weight (w_j) of chemical parameter j are calculated, as shown in Equations (5)–(7). The fourth step involves calculating the quality rating scale of chemical parameter j (q_j), based on each chemical parameter j groundwater concentration (C_j) and threshold in drinking water (S_j), as proposed by the World Health Organization (Equation (8)). In the final step, EWQI is determined by summing the products of q_j and w_j for each groundwater chemical parameter (Equation (9)).

$$X={\left[\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1n}\\ x_{21}& x_{22}& \dots & x_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ x_{m1}& x_{m2}& \dots & x_{mn}\end{array}\right]}_{m\times n}$$

(2)

$$y_{ij}={\displaystyle \frac{x_{ij}-{\left(x_{ij}\right)}_{min}}{{\left(x_{ij}\right)}_{max}-{\left(x_{ij}\right)}_{min}}}$$

(3)

$$Y=\left[\begin{array}{cccc}{y}_{11}& y_{12}& \dots & y_{1n}\\ y_{21}& y_{22}& \dots & y_{2n}\\ ⋮& ⋮& \dots & ⋮\\ y_{m1}& y_{m2}& \dots & y_{mn}\end{array}\right]$$

(4)

$$P_{ij}={\displaystyle \frac{y_{ij}}{{{\displaystyle \sum }}_{i=1}^m{y}_{ij}}}$$

(5)

$$e_j=-{\displaystyle \frac{1}{ln m}}{\displaystyle \sum }_{i=1}^m{P}_{ij}ln{P}_{ij}$$

(6)

$$w_j={\displaystyle \frac{1-e_j}{{{\displaystyle \sum }}_{j=1}^{n}1-e_j}}$$

(7)

$$q_j={\displaystyle \frac{C_j}{S_j}}\times 100$$

(8)

$$EWQI={\displaystyle \sum }_{j=1}^n{w}_j{q}_j$$

(9)

Based on the EWQI values, groundwater is classified into five categories, namely excellent (EWQI < 25), good (25 < EWQI < 50), medium (50 < EWQI < 100), poor (100 < EWQI < 200), and the extremely poor (EWQI > 200) groundwater quality category [62].

2.4. Sodium Adsorption Ratio and Sodium Percentage

Sodium adsorption ratio (SAR) and sodium percentage (%Na) were used to assess the suitability of groundwater for irrigation. SAR and %Na are crucial factors in determining whether groundwater is suitable for irrigation, given their indication of sodium’s potential impact on soil [63].

SAR and %Na are calculated as follows [64,65]:

$$SAR={\displaystyle \frac{N{a}^{+}}{\sqrt{\frac{C{a}^{2+}+M{g}^{2+}}{2}}}}$$

(10)

$$\%Na={\displaystyle \frac{N{a}^{+}}{N{a}^{+}+K^{+}+C{a}^{2+}+M{g}^{2+}}}\times 100\%$$

(11)

Depending on the SAR values, groundwater can be categorized as excellent (SAR < 10), good (10 < SAR < 18), doubtful (18 < SAR < 26), or unsuitable (SAR > 26) for irrigation. On the other hand, based on the %Na values, groundwater can be classified as excellent (%Na < 20%), good (20% < %Na < 40%), permissible (40% < %Na < 60%), doubtful (60% < %Na < 80%), or unsuitable (%Na > 80) [21].

2.5. Potential Human Exposure Assessment

A conventional (deterministic) model from the USEPA [44] was employed to assess the health impacts of fluoride and nitrate in groundwater on the population in the study area. Human health may be at risk from groundwater contaminated with fluoride and nitrate due to ingestion and dermal exposure. However, the primary route of exposure to nitrates and fluorides from groundwater is through ingestion [2,5]. Moreover, due to the absence of an established reference dose for dermal contact by the USEPA, this study did not consider this exposure pathway. Considering that fluorides and nitrates are associated with non-carcinogenic health impacts [4], this study evaluated the non-carcinogenic risk posed by these two pollutants in groundwater via the ingestion route for adults and children.

First, chronic daily intake (CDI) was calculated based on pollutant concentration (C, mg/L) and exposure parameters, such as ingestion rate (IR, L/day), exposure frequency (EF, days/years), exposure duration (ED, years), body weight (BW, kg), and averaging time (AT, days) as follows:

$$CDI={\displaystyle \frac{C\times IR\times EF\times ED}{BW\times AT}}$$

(12)

The exposure parameter values for both population groups are presented in Table S1.

The non-carcinogenic risk from the ingestion of fluoride or nitrate is quantified by the hazard quotient (HQ) (Equation (13)). Additionally, the overall non-carcinogenic risks of fluoride and nitrate are evaluated using the hazard index (HI) (Equation (14)).

$$HQ={\displaystyle \frac{CDI}{RfD}}$$

(13)

$$HI={\displaystyle \sum }HQ$$

(14)

In Equation (13), RfD denotes the reference dose (in mg/kg/day). In this study, RfD values via the ingestion route of 1.6 mg/kg/day for nitrates and 0.06 mg/kg/day for fluorides were applied, as defined by the USEPA [66].

Values of HQ and HI lower than 1 indicate that the non-carcinogenic risk is considered negligible. Conversely, if HQ and HI values are greater than 1, non-carcinogenic risk is at an unacceptable level [67,68,69].

2.6. Monte Carlo Simulation

The deterministic assessment of human health risk focuses on representing risk with a single value. In this way, the deterministic method fails to encompass the variability and uncertainty associated with human risk assessment. Therefore, assessing and identifying uncertainties and variabilities in health risks is crucial for more precise risk evaluation. Monte Carlo simulations utilize random sampling to create random data and approximate model outcomes, offering the benefit of incorporating variability and uncertainty into risk assessment to yield more insightful estimations of the likely range as opposed to single-point estimations. Monte Carlo simulation involves three steps: (1) determination of the probability distribution of parameters responsible for uncertainties, such as ingestion rate, exposure frequency, and body weight; (2) random sampling of specified values within the distribution; and (3) applying these values to deterministic health risk assessment equations [21].

The simulation was conducted using 20,000 iterations with a confidence level of 95%. The distribution types of parameters responsible for uncertainties are shown in Table S1.

2.7. Data Analysis

Descriptive statistics of groundwater parameters and normality tests of their distribution were carried out with Minitab v. 17 (Minitab Inc., State College, PA, USA). Minitab was also used to conduct Pearson correlation analysis and principal component analysis. qGIS v. 3.32 (qGIS, London, UK) was used to create the spatial distribution maps, while Origin v. 2021 (OriginLab, Northampton, MA, USA) was utilized for all graphical representations of the results. The Monte Carlo simulation was performed in Crystal Ball 11.1.24 (Oracle, Austin, TX, USA).

3. Results and Discussion

3.1. Hydrogeochemical Features of Groundwater

In this study, 19 physico-chemical parameters, including pH, T, EC, TDS, DO, HCO₃⁻, F⁻, Cl⁻, SO₄²⁻, NO₃⁻, B, SiO₂, Sr²⁺, Ca²⁺, K⁺, Li⁺, Mg²⁺, Na⁺, and PO₄³⁻, were analyzed (Table S2) and each parameter’s statistics, including mean, standard deviation (StDev), coefficient of variation (CoefVar), minimum, median, maximum, skewness, and kurtosis, are summarized in Table S3. The drinking water standards outlined by the World Health Organization [22], along with the drinking water standards in the Republic of Serbia [70], are also presented in the table. The distributions of the physico-chemical data are also presented as box plots in Figure 2.

Groundwater pH ranged from 7.3 to 9.6, averaging 7.8, implying its alkaline nature. Only one sample (no. 5) had a value outside the allowable ranges set by the WHO (6.5–8.5) and the Serbian Standard (6.8–8.5). The TDS values were in the range of 220–1860 mg/L, with a mean of 441 mg/L. All of the samples had TDS values outside the WHO permissible range of 600–1000 mg/L, with 19 samples and one sample (no. 5) having TDS values below 600 mg/L and above 1000 mg/L, respectively. Generally, water with TDS levels below 600 mg/L is typically deemed to have good palatability, while TDS levels exceeding 1000 mg/L render drinking water significantly less palatable [22]. The EC varied between 366 and 3097 μS/cm, averaging 734 μS/cm. The majority of the samples (65%) fell within the standard for drinking water of 500–1500 μS/cm. However, six samples were characterized by an EC value below the lower limit, while one sample (no. 5) had a value above the upper limit. Higher EC may arise due to human influence or elevated mineral content [31]. Considering the standards set by the Republic of Serbia, only sample no. 5 exceeded the established threshold of 2500 μS/cm.

The average concentration of anions decreased in the following order: HCO₃⁻ (484 mg/L) ≫ SO₄²⁻ (46.2 mg/L) > Cl⁻ (18.6 mg/L) > NO₃⁻ (11.1 mg/L) > F⁻ (1.0 mg/L). Among these anions, HCO₃⁻ and F⁻ showed exceedance of drinking water standards. The concentration of HCO₃⁻ surpassed the proposed WHO standard in 80% of the samples. On the other hand, the F⁻ concentration was above the permissible limits set by the WHO (1.5 mg/L) and Serbian standard (1.2 mg/L) in only one sample (no. 5), with the concentration approximately seven times higher than the allowed concentration (10 mg/L).

The level of F⁻ in drinking water is of great importance from the aspect of human health. Concentrations of F⁻ between 1.5 and 4 mg/L are associated with dental fluorosis, whereas concentrations between 4 and 10 mg/L are linked to both dental and skeletal fluorosis [25]. However, in terms of human health, even F⁻ concentrations that are too low are considered undesirable. In particular, chronic exposure to drinking water with F⁻ concentrations below 0.5 mg/L is associated with dental caries. In the study area, concentrations of F⁻ ranged from 0.015 mg/L to 10 mg/L, with 40% of the groundwater samples exhibited F⁻ levels below 0.5 mg/L, suggesting that the inhabitants of the area are at possible risk of developing dental caries. For comparison, F⁻ concentration in groundwater in the Indian state of Uttar Pradesh [71], as well as in Northern Punjab in Pakistan [72], averaged 0.97 mg/L and 0.90 mg/L, respectively, similar to the findings reported in this study.

The spatial distribution of F⁻ concentration in groundwater is shown in Figure 3a. It can be seen that the majority of the area is characterized by F⁻ concentrations in the range of 0.5–1.5 mg/L, indicating that groundwater is safe from the aspects of human health. Only a small portion throughout the area exhibited an F⁻ below 0.5 mg/L, whereas the central part was of the poorest water quality, with concentrations of F⁻ ions above 1.5 mg/L, reaching 10 mg/L.

It is observed that the concentration of NO₃⁻ was within the permissible limit of 50 mg/L set by both standards in all samples, ranging from 1.4 mg/L to 36 mg/L. However, natural NO₃⁻ concentrations in groundwater are generally believed to be below 5 mg/L, with concentrations above this indicating human-induced influences [2]. Concentrations of NO₃⁻ above 5 mg/L were characterized for the majority of the samples (13 out of 20 samples), suggesting anthropogenic influence in the study area. Similar impacts from human activities on groundwater quality have been documented in other regions, such as the Yellow River Plain in China, where NO₃⁻ levels reached as high as 706.4 mg/L [3], and the River Vomano Plain in Italy, with concentrations up to 120 mg/L [73]. Both regions experience significant agricultural activity, which is the main factor behind the elevated NO₃⁻ levels observed in the groundwater. This comparison highlights the widespread influence of agricultural activities on NO₃⁻ contamination.

The spatial distribution of NO₃⁻ concentration is depicted in Figure 3b. It was observed that the majority of the area exhibited NO₃⁻ concentration above 5 mg/L, implying an anthropogenic influence on groundwater quality. Groundwater of the worst quality with regard to NO₃⁻ concentration is seen in the northeastern and central portions of the area, exhibiting NO₃⁻ levels of more than 20 mg/L.

The average concentration of cations decreased in the order of Ca²⁺ (90 mg/L) ≫ Na⁺ (62.1 mg/L) > Mg²⁺ (21.8 mg/L) > K⁺ (3.0 mg/L) > Li⁺ (0.41 mg/L). Among these cations, Ca²⁺, Mg²⁺, and Na⁺ exceeded the drinking water standards. The concentration of Ca²⁺ exhibited the greatest variance from the specified WHO limit, surpassing the allowable threshold in 12 out of 20 samples. On the other hand, all samples met the Serbian standard for Ca²⁺ of 200 mg/L. However, only one sample exceeded the limits for drinking water regarding Na⁺ and Mg²⁺ concentrations, which are the same in both standards. More specifically, an extremely high Na⁺ concentration was observed in sample no. 5, reaching 900 mg/L.

On the other hand, Mg²⁺ concentration showed a slight exceedance of the established drinking water limits, reaching 61 mg/L. The presence of abundant lithium ores in the study area results in exceptionally high Li⁺ levels in the groundwater. The concentration of Li⁺ ranged from 0.07 to 5.8 mg/L, averaging 0.41 mg/L. The maximum Li⁺ concentration was again recorded in sample no. 5. There are no guidelines on the acceptable levels of lithium in drinking water, owing to the undetermined effects of lithium on human health [74]. However, a non-regulatory screening level for lithium of 60 μg/L [75], based on the premise that drinking water is the sole source of daily lithium consumption, was exceeded in all samples several times.

Sample no. 5 deviated most significantly from the established standards, exhibiting exceptionally elevated values for pH, EC, and TDS, as well as concentrations of F⁻, Li⁺, and Na⁺. Sample no. 5 was taken from a well exceeding 520 m deep, specifically drilled to explore deposits of lithium. Therefore, the elevated concentrations of parameters in sample no. 5 can be attributed to the geological characteristics of the terrain, which explain why it was not classified as an outlier.

The Piper [76] diagram was employed to determine the hydrogeochemical types present in the groundwater (Figure 4a). The majority of the samples were plotted in zone A of the cation triangle, indicating calcium groundwater type. As seen from the anion triangle, all samples were plotted in zone E, illustrating the bicarbonate groundwater type. It was observed that the sample with elevated F⁻ concentration was of Na-K-HCO₃ hydrogeochemical type (Figure 4a,b), falling into zone 3, as expected due to its high Na⁺ and HCO₃⁻ concentration. This aligns with the hydrogeochemical characteristics of elevated F⁻ groundwater observed in other regions [16,20]. However, the majority of the samples (95%) fell within zone 1 in the diamond, revealing Ca-Mg-HCO₃ as the predominant hydrogeochemical type. The predominance of Ca-Mg-HCO₃ indicates that water–rock interaction influences the chemical composition of groundwater. Consequently, the elevated F⁻ levels in groundwater primarily originate from the dissolution of fluorine-bearing minerals rather than from human activities.

Furthermore, the Gibbs [77] diagram was employed to illustrate the impact of the three primary mechanisms (atmospheric precipitation, rock weathering, and the evaporation-crystallization process) on groundwater chemistry. As evidenced by Figure 5, groundwater samples of Ca-Mg-HCO₃ type fell within the rock weathering zone, indicating that water–rock interactions govern the chemistry of groundwater in the study area. However, the Na-K-HCO₃ type showed a modest increase in trend toward areas dominated by evaporation. Limestone, sandstone, and claystone predominantly cover the area. The weathering of these rocks with their interaction with groundwater leads to the release of major ions such as Na⁺, Ca²⁺, Mg²⁺, and HCO₃⁻ in the groundwater.

3.2. Quality of Groundwater for Drinking and Irrigation

The suitability of groundwater for drinking is frequently evaluated using the entropy-weighted water quality index (EWQI). To calculate EWQI, this study took into account groundwater parameters that have been prescribed as health-based or taste-based maximum permissible levels in drinking water by the WHO, including pH, EC, HCO₃⁻, TDS, F⁻, Cl⁻, SO₄²⁻, NO₃⁻, PO₄³⁻, B, Ca²⁺, K⁺, Mg²⁺, and Na⁺.

The water supply in the Kolubara district is mostly based on groundwater reservoirs (55%) [78]. Therefore, it is extremely important to assess the quality of groundwater for drinking. The EWQI ranged between 12.0 and 572, with a mean value of 50.4. Excellent groundwater quality (EWQI < 25) was observed in 60% of the samples, whereas 35% of the samples exhibited good groundwater quality (25 < EWQI < 50). These waters are considered suitable for drinking. None of the groundwater samples fell into the medium (50 < EWQI < 100) and poor (100 < EWQI < 200) quality categories. However, one sample (no. 5) was classified as extremely poor quality (EWQI > 200), exhibiting a maximum EWQI value of 572. These results can be attributed to extremely elevated concentrations of F⁻, HCO₃⁻, B, and Na⁺ in sample no. 5. Given the EWQI results, the groundwater in the study area is exhibiting excellent to good quality for drinking, with the exception of sample no. 5. It is advised to exclude this sample location from drinking water supply, considering that groundwater with that quality is both unpleasant in taste, due to high levels of HCO₃⁻ and Na⁺, and harmful to health, given the elevated F⁻ and B concentrations.

The spatial distribution of EWQI in the study area is depicted in Figure 6. Most of the area is characterized by excellent to good groundwater quality (62.1% of the area). Medium groundwater quality was observed in 36.3% of the area, mostly in the southeastern parts and extending to the northeast. These waters are appropriate for domestic utilization. Nonetheless, pre-treatment is advisable prior to their use in households [21]. A small portion of the area (1.6%) exhibited EWQI greater than 100 (poor and extremely poor quality categories), representing waters unsuitable for drinking. The worst quality groundwater was observed in the central part of the study area in close proximity to sample no. 5.

The population of the investigated area is extensively engaged in agriculture. Increasing the yield of agricultural crops is achieved, among other things, by irrigation. Although the primary source for this purpose is river water from the Kolubara River basin, groundwater accounts for around 40% of the irrigation supply [55]. Therefore, it is important to know whether groundwater is suitable for irrigation. The quality of groundwater for irrigation was determined by %Na and SAR indices, along with the Wilcox [79] diagram.

Sodium content is one of the most important parameters for irrigation water use, considering high sodium content in irrigation water reduces soil permeability [80]. Table S4 shows the groundwater classification results based on %Na and SAR.

In this study, %Na ranged between 4.0% and 97.6%, averaging 15.1%. A total of 95% of samples showed %Na values lower than 40%, falling into excellent to good quality categories. Generally, groundwater exhibited %Na values below 30%. Only one sample (no. 5) showed an extreme %Na value of 97.6%, given its elevated Na⁺ concentration of 900 mg/L.

The SAR ranged from 0.14 to 61.2, with an average value of 3.5, indicating that 95% of the samples had low sodium hazards and were excellent for irrigation. As expected, the remaining sample (no. 5) falls into an unsuitable category with a high sodium hazard, characterized by the highest SAR value.

Based on the Wilcox diagram, shown in Figure S1, the majority of the samples (19 out of 20) were plotted in excellent (75%) and good (20%) zones for irrigation. Sample no. 5, also due to its high Na⁺ concentration, fell into the unsuitable category.

Given the irrigation water quality results, groundwater in the study area can be considered suitable for agricultural use. However, groundwater from sample no. 5 should be excluded from this consideration, as elevated Na⁺ levels in water can adversely affect the structure of the soil, consequently inhibiting the development of plants.

3.3. Source Apportionment of Groundwater Fluoride and Nitrate

To determine the relationship between groundwater parameters and to evaluate their possible origins, principal component analysis (PCA) coupled with Spearman correlation analysis was conducted. Spearman correlation analysis was chosen, considering the data did not follow a normal distribution. Furthermore, to achieve better data interpretation, varimax rotation with Kaiser normalization was employed for PCA. Table S5 shows the PCA results, comprising the extracted factors’ loadings, their eigenvalues, and the percent of overall variance explained, while Figure 7 illustrates the Spearman correlation matrix.

Five principal components (PC) were extracted with eigenvalues greater than 1, explaining 89.5% of the total variance.

The first principal component (PC1), explaining 50.0% of the overall variation, had strong positive loadings of TDS, EC, F⁻, HCO₃⁻, B, Li⁺, and Na⁺ (Table S5), suggesting their similar source. Generally, the major cations and anions are associated with the dissolution of carbonate and silicate minerals, reflecting water–rock interactions [81]. The second principal component (PC2), which explained 20.2% of the total variance, was strongly loaded with SO₄²⁻, Ca²⁺, and Mg²⁺. These ions’ contribution to PC2 is attributable to gypsum dissolution and cation-exchange interactions at the soil–water boundary [82]. Moreover, Quaternary sediments of the study area, composed of limestone, sandstone, clay, and siltstone, serve as a potential source of HCO₃⁻, Ca²⁺, Mg²⁺, and Na⁺ to groundwater [83]. Therefore, PC1 and PC2 can be explained as geogenic sources. Additionally, as per the Spearman correlation matrix (Figure 7), a significant (p < 0.05) positive correlation was found between TDS and F⁻, SO₄²⁻, Na⁺, Li⁺, Mg²⁺, and Ca²⁺. The prolonged circulation of groundwater within an aquifer results in the dissolution of anions and cations from rock formations into groundwater, which is reflected in the significant correlation observed between TDS and these ions. Moreover, HCO₃⁻ and SO₄²⁻ both showed significant positive correlations with Ca²⁺, Li⁺, and Mg²⁺. These results imply that groundwater in the study area is mainly influenced by natural processes.

The dissolution of minerals containing fluorine, such as fluorite (CaF₂), fluorapatite (Ca₅(PO₄)₃F), and villianmite (NaF), is the primary source of F⁻ in groundwater [27]. This process is primarily influenced by the composition of the bedrock material, the duration of water–rock interaction, pH, and the ion exchange processes in the groundwater [72]. In this study, F⁻ concentrations were generally low and within the acceptable limit of 1.5 mg/L, except for sample no. 5, confirming natural processes as the dominant source of F⁻ in the groundwater of the study area. Moreover, lithium, which is of geogenic origin in the study area, exhibited a significant positive correlation with F⁻ (r = 0.64), indicating that both F⁻ and lithium share the same source.

The dissolution of fluorine-bearing minerals and the mobilization of fluoride are significantly driven by pH. More specifically, alkaline pH conditions promote the release of F⁻ through ion exchange with OH⁻ ions [4]. In this study, F⁻ showed weak positive correlation with pH (r = 0.21). However, sample no. 5, with the elevated F⁻ concentration (10 mg/L), had an alkaline pH (9.6), suggesting that an alkaline environment facilitates the mobilization of F⁻ in the study area. Moreover, F⁻ demonstrated significant positive correlation with HCO₃⁻ (r = 0.53), Na⁺ (r = 0.52), and Li⁺ (r = 0.64). The competitive adsorption of HCO₃⁻ impacts F⁻ concentrations in groundwater [16]. Bicarbonate may raise F⁻ levels by competing with F⁻ for adsorption sites, leading to the desorption of F⁻ from sediments into groundwater. The positive linkage between F⁻ and HCO₃⁻ suggests competitive adsorption as a mobilization mechanism of F⁻. Additionally, the significant association between Na⁺ and Li⁺ with F⁻ indicates that higher Na⁺ and Li⁺ levels enhance the transport of F⁻ in solutions with an alkaline pH [4]. High F⁻ concentration in sample no. 5 was associated with its alkaline pH of 9.6 and high Na⁺ (900 mg/L) and Li⁺ (5.8 mg/L) content. These conditions, along with the low Ca²⁺ level (12 mg/L), demonstrate that cation exchange could have contributed to the elevated F⁻ levels in the study area’s groundwater. Moreover, the natural source of F⁻ in groundwater in the investigated area was confirmed by the PCA results, where F⁻, along with TDS, HCO₃⁻, Li⁺, and Na⁺, demonstrated high positive loadings on PC1.

The third principal component (PC3) explained 8.0% of the total variation. PC3 exhibited a strong positive loading of NO₃⁻ only, possibly pointing to human influence on groundwater quality through intensive agricultural activities. This is particularly due to the use of KNO₃ fertilizers, which are highly water-soluble, or animal manure [21]. Hence, PC3 can be explained as the agricultural source.

Nitrates in groundwater systems are indicative of anthropogenic contamination, given the absence of substantial geogenic sources [5]. It is considered that the concentration of NO₃⁻ in groundwater is naturally lower than 5 mg/L, and that concentrations higher than that point to anthropogenic influence [2]. Agricultural practices, including an overuse of artificial fertilizers and pesticides, are the main activities responsible for elevated levels of NO₃⁻ in groundwater [17]. Although the concentration of NO₃⁻ in the investigated area was within the permissible limit prescribed by the WHO, 65% of the samples were characterized by a concentration higher than 5 mg/L, which indicates an anthropogenic agricultural source. This is expected considering that the inhabitants of the study area are extensively engaged in agriculture.

A negative correlation between F⁻ and NO₃⁻ is often interpreted as a sign of human activities due to the natural groundwater lacking NO₃⁻ [16]. Spearman correlation analysis revealed a moderate negative correlation between F⁻ and NO₃⁻ (r = −0.41), which implies that these two ions originate from different sources. The geogenic lithium in the area under study also showed a negative correlation with NO₃⁻ (r = −0.30), implying that NO₃⁻ originates from anthropogenic sources. Moreover, the third principal component (PC3) showed the highest positive loading for NO₃⁻, while F⁻ demonstrated negative loading, further suggesting their different sources. Additionally, pH was negatively correlated with NO₃⁻, which can be explained by denitrification in alkaline environments [72]. Consequently, the denitrification process may be responsible for maintaining NO₃⁻ levels within the permissible limits in the groundwater of the study area.

Apart from the three principal components mentioned above, two more principal components with eigenvalues greater than one were obtained by PCA. The fourth (PC4) and fifth principal components (PC5) explained 5.8% and 5.4% of the total variance, respectively. High positive loadings of K⁺ and Cl⁻ were characterized for PC4 and PC5, respectively. The primary sources of K⁺ and Cl⁻ in groundwater are domestic and municipal wastewater and sewage [84]. Infiltration or leakage of municipal and domestic wastewater can result in K⁺ and Cl⁻ reaching groundwater. Hence, PC4 and PC5 can be characterized as municipal and domestic wastewater sources. According to the Statistical Office of the Republic of Serbia’s 2022 report, approximately 37% of households in the Kolubara district had access to the public sewage system, with the majority of the remaining households using septic tanks [85].

3.4. Potential Human Exposure to Fluoride and Nitrate from Groundwater

The results of the non-carcinogenic health risk assessment for adults and children are presented in

Table 1. The results of non-carcinogenic health risk assessment for adults and children in the study area.

Health Risk Index		HQ		HI
		NO3−	F−	Total
Adults	Mean	0.25	0.59	0.84
	Max	0.8	5.95	5.99
	Min	0.03	0.01	0.19
Children	Mean	0.36	0.86	1.23
	Max	1.17	8.67	8.72
	Min	0.05	0.01	0.28

. Regarding adults, HQ of NO₃⁻ ranged from 0.03 to 0.80, averaging 0.25, while HQ of F⁻ was in the range 0.01–5.95, averaging 0.59. Regarding children, HQ of NO₃⁻ ranged between 0.05 and 1.17, averaging 0.36, while HQ of F⁻ was in the range of 0.01–8.67, averaging 0.86. It was observed that F⁻ was characterized by higher non-carcinogenic risk in both population groups, making it a more contributing element to risk. Additionally, children were characterized by higher HI values for both F⁻ and NO₃⁻. Total non-carcinogenic risk, expressed as the sum of non-carcinogenic risk from F⁻ and non-carcinogenic risk from NO₃⁻, ranged from 0.19 to 5.99 for adults and from 0.28 to 8.72 for children. In total, 85% and 65% of the samples were within the permissible limit of one for adults and children, respectively. These findings suggest that children were more prone to non-carcinogenic risk compared to adults, which can be attributed to children’s lower body weight and exposure duration [1,21].

The spatial distribution of non-carcinogenic risk in both exposure groups is shown in Figure 8. The western portion of the study area exhibits the lowest risk, whereas the eastern portions are at higher risk. The highest risk is observed in the central part of the area, with one hotspot detected around sampling point no. 5, the one characterized by an extremely elevated concentration of F⁻ (10 mg/L). Moreover, the majority of the study area exhibited HI values above the limit of one (health risk zone) for children (63.7%), whereas that area was significantly smaller for adults (13.1%) (Figure S2). These results further confirm that children are at greater non-carcinogenic risk than adults and are of great concern since exposure to F⁻ and NO₃⁻ in drinking water can cause various health problems, including skeletal fluorosis and methemoglobinemia [14,23]. Therefore, special attention should be paid to mitigating health risks in the study area.

Figure 9 depicts the probability distributions of the hazard quotients of F⁻ and NO₃⁻ for both adults and children, along with the probability distributions of non-carcinogenic health risks for adults and children obtained by Monte Carlo simulation. Statistical information regarding distributions, including minimum, maximum, mean, median, standard deviation, skewness, kurtosis, 10th percentile, and 90th percentile, is presented in Table S6. Since sample no. 5 drastically differed from the rest of the samples, and as such, it would negatively affect the results of the probabilistic distribution, making them inadequate or inaccurate. In order not to overestimate the risk, sample no. 5 was excluded from the Monte Carlo simulation.

As seen in Figure 9, the HQ values for both F⁻ and NO₃⁻ did not exceed the threshold of 1 with regard to both adults (HQa) and children (HQc). The following range of values is observed: HQc_F⁻ > HQc_NO₃⁻ > HQa_F⁻ > HQa_NO₃⁻, implying that children are at greater risk and that F⁻ is the main health risk contributor. Moreover, the average HQ value of NO₃⁻ and F⁻ for children was approximately 46% and 48% higher, respectively, than those for adults. Considering that F⁻ is of geogenic origin, geogenic contamination of groundwater in the study area is a major concern for human health.

Furthermore, the probabilistic distribution of HIa ranged from 0.30 to 1.02, averaging 0.58 (Table S6). The value of HIa, even at the 90th percentile, was below the permissible limit of 1. Nonetheless, it was observed that there exists a 0.46% probability that HIa will exceed the specified limit (Figure 9). On the other hand, HIc values were much higher, ranging from 0.45 to 1.45, with a mean value of 0.84 (Table S6). The value of HIc at the 10th percentile was lower than the limit of 1, whereas at the 90th percentile, it equaled 1. Compared with adults, the probability of HIc surpassing the threshold of 1 was approximately 29-fold higher, amounting to 13.3% (Figure 9). These results indicate that although the concentrations of NO₃⁻ and F⁻ are below the permissible limits in drinking water, health risks still exist. The simultaneous presence of NO₃⁻ and F⁻, despite each being below their respective limits, resulted in combined effects that led to elevated overall health risks. Hence, conducting a health risk assessment remains crucial, even for water deemed clean, to precisely assess the potential health implications of its consumption.

The Monte Carlo simulation results were in agreement with the deterministic health risk assessment results, confirming that children were at much greater risk than adults. However, compared to the deterministic model results, the HIa and HIc threshold exceedances were significantly lower in the probabilistic model. This can be attributed to the fact that the Monte Carlo model does not use a single fixed value for health risk parameters; instead, it employs an appropriate distribution, allowing for a more accurate estimation of risk without overestimation.

3.5. Public Health Protection Strategies and Future Research Perspectives

To summarize, groundwater in the Kolubara area is not polluted with F⁻ and NO₃⁻. However, the health risk assessment indicated a potential risk of non-carcinogenic diseases, particularly among children. In this context, several recommendations were proposed to mitigate health risk in the area, as outlined below.

(a)

Primarily, ongoing groundwater monitoring is crucial for safeguarding human health. This approach allows for the early detection of contaminants, facilitates timely intervention, and ensures that water quality remains within safe limits.

(b)

To lower F⁻ concentrations, in-situ approaches such as rainwater recharge through existing wells, rainwater harvesting, or artificial recharge can be employed [86]. These approaches effectively lower fluoride concentrations in groundwater by utilizing the dilution effect.

(c)

Effective management of nitrogen fertilizers is crucial to minimize the leaching of nitrogen into groundwater in agricultural regions. To mitigate environmental impact, farmers are suggested to lower the reliance on nitrogen fertilizers and utilize organic alternatives [87,88].

To fully determine the effect of agricultural activities on groundwater quality, future work should consider investigating seasonal variations of NO₃⁻ in groundwater. Additionally, establishing a numerical model could enhance the simulation of NO₃⁻ dynamics and provide insights into the time required for NO₃⁻ to move from the soil to groundwater.

4. Conclusions

This study investigated the hydrogeochemistry and quality of groundwater in the lithium ore deposit basin, explored the distribution and origin of F⁻ and NO₃⁻ in groundwater, and utilized deterministic and probabilistic models to evaluate non-carcinogenic health risks associated with F⁻ and NO₃⁻ from the ingestion exposure route, resulting in the following conclusions:

(1)

Groundwater hydrogeochemistry is governed by the dissolution of rocks, with Ca-Mg-HCO₃ being the predominant groundwater type.

(2)

The EWQI results suggested that groundwater was suitable for drinking, with most of the samples (95%) falling into the categories of excellent to good water quality. Water unfit for drinking was observed in the central regions of the area, while the rest of the area was characterized by groundwater suitable for drinking. Moreover, irrigation indices (SAR and %Na) revealed that groundwater has a low sodium hazard, making it suitable for irrigation.

(3)

Concentration of F⁻ exceeded the WHO drinking water guideline in 5% of samples. Its release into groundwater is primarily governed by natural processes such as pH, competitive adsorption, and cation exchange. Concentration of NO₃⁻ was within the permissible WHO threshold for drinking in all samples. However, 65% of samples exhibited NO₃⁻ concentration above 5 mg/L, revealing an anthropogenic influence.

(4)

Groundwater in the study area, although suitable for consumption, poses potential health risks to humans. Deterministic and probabilistic health risk assessments revealed that children face higher non-carcinogenic risk. Moreover, F⁻ from natural sources has a dominant impact on human health. These results suggest that groundwater in the study area should be continuously monitored and can serve as guidance to decision-makers to pay attention to people’s health, with special emphasis on children.