Studying the applicability of electrodialysis technology to treat groundwater in CKDu regions, Sri Lanka: operational parameter optimization of the pilot project

Groundwater is considered one of the most significant natural resources of drinking water in many parts of the world. Globally, 30% of the total freshwater demand is met by groundwater, which provides almost half of all the drinking water requirements worldwide. Many people in arid regions in Sri Lanka also consume groundwater as their main drinking water source, with 87% of them directly using groundwater from shallow-dug wells or deep tube wells (Makehelwala et al. 2019).

The quality of groundwater is mainly determined by location-specific geochemistry. Therefore, groundwater quality must be carefully examined to maximize utilization and sustainability. Chronic kidney disease of unknown etiology (CKDu) hotspots are geographically scattered and overlap in rural dry regions, especially in the North Central Province (NCP) and its adjacent districts where groundwater is the main drinking water source (Wickramarathna et al. 2017).

CKDu is progressively found in tropical regions and has been recognized as a major global health problem in recent years. In Sri Lanka, CKDu has been reported over the last three decades since the initial incidents were reported in the 1990s. The cause of CKDu is still unknown; however, prior investigations have proposed numerous hypotheses. Most of these hypotheses are directly or indirectly related to the water consumption patterns of the patients. Some studies have shown that groundwater quality has a direct effect on the emergence of CKDu. The pH, alkalinity, total hardness, electrical conductivity (EC), calcium, magnesium fluoride, chloride, phosphate, and sulfate of the wells used by CKDu patients were significantly high. The dissolution of aquifer minerals and a higher rate of evaporation were the main reasons for the higher ion contents in water in the dry regions of Sri Lanka. These excessive ions need to be removed in these regions to provide safe water, which may impact CKDu incidence in Sri Lanka (Balasubramanya et al. 2020).

Physical, chemical, and electrochemical drinking water treatment systems have been proposed for impoverished areas in Sri Lanka. In recent decades, membrane technologies have gained popularity with the most widely utilized rural water treatment method in Sri Lanka being reverse osmosis (RO) due to its minimal space requirements, flexible installations, and ability to manage diverse influent water quality (Ketharani et al. 2022). Despite these benefits, RO plants have drawbacks such as high energy consumption, membrane fouling, inadequate water recovery, and the generation of highly concentrated waste brine (Skuse et al. 2021). Therefore, people are researching the electrodialysis (ED) technique for groundwater treatment. ED desalination is cost-effective and widely utilized in solution treatment where ion-permeable membranes separate dissolved ions under an electrical potential gradient in the ED process (Al-Amshawee et al. 2020).

In some countries, ED is used to treat drinking water. Moroccan authorities have implemented ED for groundwater treatment, where fluoride concentrations vary seasonally. This study indicated that predetermined operational parameters can readily achieve the desired drinking water quality. In addition, pilot plants for groundwater treatment have been operated using conventional ED treatment in Morocco, India, and South Africa (Malalgama et al. 2022). However, traditional ED technology cannot selectively remove ions. Therefore, some essential ions are also removed during the treatment. It was found that a few selected ions need to be treated to obtain quality water for drinking purposes in Sri Lanka. Hence, traditional ED was modified to selectively remove ions with changes in operational parameters in the system.

This paper reports on parameter optimization and performance monitoring of the first groundwater selective ED desalination plant installed in the dry zone of Sri Lanka, where the concentration of a few ions, such as those that cause hardness, alkalinity, EC, and fluoride, exceeds the SLS 614:2013 (Sri Lanka Standard 614:2013) guideline limits.

Systematic water quality monitoring was conducted in Kahatagasdigiliya to understand groundwater quality issues to be addressed during the ED plant installation. Groundwater samples were collected from two drilled boreholes (BH1: 8.420624, 80.699786 and BH2: 8.420669, 80.699713) which the villagers have been using for their daily water requirements. Samples were collected for 5 years from March 2013 to February 2018. Summarized water quality characteristics of these two boreholes are shown in Table 1.

Table 1

Major water quality characteristics of samples obtained from two boreholes at the study location

Water parameter .	Borehole 1 .			Borehole 2 .			SLS 614:2013 .
	Min .	Max .	Avg. .	Min .	Max .	Avg. .	Stand. .
Color	0	20	7.05	2	16	6.66	15
Turbidity (NTU)	0.01	11.3	0.90	0.02	1.63	0.47	2
pH	6.43	7.54	6.80	6.32	7.76	6.74	6.5–8.5
EC (μs/cm)	610	868	725.71	600	798	684.10	500
Alkalinity (as CaCO3) (mg/L)	168	390	260.75	190	300	239.93	200
Hardness (as CaCO3) (mg/L)	160	330	252.40	150	344	235.92	250
Fluoride (as F−) (mg/L)	0.64	1.52	1.09	0.59	1.38	1.01	1.0
Chloride (as Cl−) (mg/L)	12	100	64.24	0	116	63.80	250
Sulfate (as ⁠) (mg/L)	0	19	11.51	2	18	10.10	250
Phosphate (as ⁠)(mg/L)	0.27	1.78	0.81	0.36	1.65	0.81	2.0
Total iron (as Fe3+) (mg/L)	0	0.7	0.05	0	0.08	0.024	0.3
Nitrate (as ⁠) (mg/L)	0.2	5.7	1.45	0.2	7.48	1.53	10.0
Nitrite (as ⁠) (mg/L)	0.001	0.073	0.0085	0.0005	0.152	0.015	3.0

Water parameter .	Borehole 1 .			Borehole 2 .			SLS 614:2013 .
	Min .	Max .	Avg. .	Min .	Max .	Avg. .	Stand. .
Color	0	20	7.05	2	16	6.66	15
Turbidity (NTU)	0.01	11.3	0.90	0.02	1.63	0.47	2
pH	6.43	7.54	6.80	6.32	7.76	6.74	6.5–8.5
EC (μs/cm)	610	868	725.71	600	798	684.10	500
Alkalinity (as CaCO3) (mg/L)	168	390	260.75	190	300	239.93	200
Hardness (as CaCO3) (mg/L)	160	330	252.40	150	344	235.92	250
Fluoride (as F−) (mg/L)	0.64	1.52	1.09	0.59	1.38	1.01	1.0
Chloride (as Cl−) (mg/L)	12	100	64.24	0	116	63.80	250
Sulfate (as ⁠) (mg/L)	0	19	11.51	2	18	10.10	250
Phosphate (as ⁠)(mg/L)	0.27	1.78	0.81	0.36	1.65	0.81	2.0
Total iron (as Fe3+) (mg/L)	0	0.7	0.05	0	0.08	0.024	0.3
Nitrate (as ⁠) (mg/L)	0.2	5.7	1.45	0.2	7.48	1.53	10.0
Nitrite (as ⁠) (mg/L)	0.001	0.073	0.0085	0.0005	0.152	0.015	3.0

View Large

The targeted electrodialysis (TED) stack contains 300 pairs of heterogeneous ion-exchange membranes. NEO-SEPTA CMX is used as the cation exchange membrane and NEO-SEPTA ACS is used as the anion exchange membrane. The surface area of each membrane is 1.28 m² (i.e., 1.60 × 0.80 m). They are separated with polypropylene spacers, which are spoused to act as flow drivers during the operation. Titanium-coated ruthenium electrodes (two metal plates) are arranged at both ends of the TED stack to act as anode and cathode.

As the first step, groundwater is pumped from deep wells (BH1 and BH2) into the collection tank (item 15, Figure 2). Then the water is filtered through a rapid sand filter with particle sizes of sand ranging from 0.5 to 0.85 mm (item 8, Figure 2), to remove suspended particles and reduce turbidity. Next, the water is pumped (item 6, Figure 2) across a security filtering device (item 7, Figure 2) to intercept small particles of suspended solids to ensure that the turbidity of the inlet water remains less than 1 Nephelometric Turbidityunit (NTU). After that, the water is desalinated using the TED stack (item 1, Figure 2). The treated water flows directly into the storage tank (item 13, Figure 2) as drinking water. Simultaneously, the concentrated water accumulates in a separate tank called the ‘concentrated water storage tank’ (item 14, Figure 2), and this water is recirculated through the TED stack continuously to minimize water wastage. Acid is added at set intervals (every 4 h) to the electrode rinse water, which is conveyed to the electrode from the concentrated water tank (item14, Figure 2). Periodically, water from the concentrated water storage tank is discharged for irrigation while keeping the liquid level steady. The TED system is operated at a constant voltage by adjusting the appropriate voltage through the power cabinet (item 11, Figure 2). The flow of each pipeline is regulated by the valve operator system (item 2, Figure 2).

Orthogonal experiments were performed to evaluate the impacts of the applied voltage and applied flow rate on hardness and F⁻ removal in TED performance. Three different voltages 45, 65, and 95 V were utilized. The flow rate also varied as 8, 12, and 16 m³/h during the experiments.

The TED equipment requires an inverted operation every day to minimize fouling in membranes. When the polarity is reversed, the dilute and concentrate chambers are also reversed. Before executing polarity reversal on the system, the TED should be shut down. Then the valves needed to be adjusted and the TED's power switched to ‘Forward Running’ or ‘Negative Running’ on the touch operation screen of the power cabinet (item 11, Figure 2). In this study, experiments with a 72 h ED operation with different numbers of electrode polarity reversals (as 1, 2, 3, 5, 10, 15, 20, 30, 50, 100, and 150) were applied to identify the optimum treated water production in the TED system.

Physicochemical analyses were carried out to characterize the water samples. pH and EC of the water samples were analyzed using the Thermo Scientific Orion Star A325 Multiparameter meter. Color, nitrate, nitrogen dioxide ⁠, phosphate, sulfate, fluoride, and total iron were measured using the spectroscopic method (DR 500). Turbidity was measured using the Lovibond TB 300 IR meter. Chloride, alkalinity, and hardness were measured using titrimetric methods as stated in the American Public Health Association (APHA) guidelines.

The general characteristics of samples collected from the boreholes (BH1 and BH2) are listed in Table 1. pH is considered one of the essential parameters in assessing water quality, as it regulates the quantity and chemical structure of dissolved organic and inorganic substances in groundwater (Sadat-Noori et al. 2014). As recommended by the SLS 614:2013 water quality guidelines, the pH of drinking water should range from 6.5 to 8.5. 92.8% of samples from both boreholes (BH1 and BH2) remained within this permissible level throughout the study period. Sulfate and chloride are crucial indicators for evaluating water quality during water analysis, as they are considered indicators of anthropogenic water pollution. Excessive presence of sulfate and chloride in drinking water can also cause changes in taste (Sadat-Noori et al. 2014). The concentration of sulfate and chloride was far below the permissible level in both BH1 and BH2 which alleviates concerns about their removal during water treatment at the location. Phosphate, nitrate, and nitrogen dioxide most often pollute groundwater due to the use of fertilizers (Piyathilake et al. 2022). In this area, remarkably low levels of phosphate, nitrate, and nitrogen dioxide were observed compared to another agricultural area in Sri Lanka (Young et al. 2010) by reducing the focus on removing these ions through the proposed drinking water treatment process. The turbidity of samples ranged from 0.01 mg/L to 11.3 NTU and 85.7% of samples were within the acceptable range. Color also followed the same pattern as turbidity, possibly due to inter-monsoonal variations in the region (Perera et al. 2013). But the removal of turbidity and color by pretreatment is the common practice.

EC is the ability of a material to conduct an electric current. The EC for the tested samples ranged from 600 to 868 μs/cm with an average of 723.12 μs/cm. In addition, EC was relatively high from 2014 to 2017. This could be due to the proximity of the study location to the arid climatic zone. The ions that leached into the groundwater from rock–water interactions may be further concentrated due to the high evaporation that prevails in the dry climatic zone, resulting in high EC values (Cooray et al. 2019). Higher EC indicates a higher concentration of ions, which leads to greater salinity or dissolved solids, which should be reduced through the treatment process.

The alkalinity of water is caused mainly by the presence of ions such as ⁠, or OH⁻ in groundwater. The alkalinity of samples was found to be in the range of 168 to 390 mg/L, which is marginally low or above the permissible limit of 200 mg/L as per the guideline of SLS 614:2013. This is common in the groundwater of the dry zone of Sri Lanka where our study area was located, and it originates due to silicate and calcite mineral weathering and the ion-exchange process in the region (Indika et al. 2022). Though there is no health concern related to alkalinity, it may contribute to scaling, which could reduce the membrane lifespan.

Total hardness refers to the amount of dissolved calcium and magnesium in water, equivalent to CaCO₃. Higher hardness gives a bitter taste to drinking water and causes scaling on pipelines and cooking utensils, which are considered issues related to high hardness (Indika et al. 2022). Hardness (as CaCO₃) surpassed the SLS 614:2013 recommendation limit (250 mg/L) in half the samples during the study period. According to Figure 4(c), hardness was high in the first half of the year and reached its peak in June, then decreased in the second half showing a cyclical pattern in hardness variation on sites, possibly due to the seasonal variation caused by inter-monsoonal rains. This is further confirmed by similar hardness fluctuation patterns observed in both studied boreholes. Most of these groundwater samples were categorized as very hard (150–300 mg/L CaCO₃) (Ras & Ghizellaoui 2012), signaling critical health issues (WHO 2011) by highlighting the importance of reducing hardness in groundwater before consumption.

Fluoride is an essential element for human health. However, excessive fluoride consumption leads to severe health issues, including dental or skeletal fluorosis disease, while a lack of fluoride causes dental decay (Bhatnagar et al. 2011). Therefore, the World health organization (WHO) recommends the healthy fluoride range for drinking purposes as 0.5–1.5 mg/L (Srivastava & Flora 2020) and the SLS 614:2013 standard has been established as 1 mg/L. The fluoride levels varied greatly in the studied locations throughout the study period, as depicted in Figure 4(d). Initially, most samples had fluoride levels below the SLS 614:2013 guideline (1 mg/L); however, they began to rise in January 2014 and eventually exceeded the SLS 614:2013 standard value, but the trend was inverted and began to fall again in July 2016. The average fluoride value exceeds SLS 614:2013 guidelines during the studied period, as depicted in Table 1. Most groundwater fluoride comes from rock and soil dissolution or volcanic particle weathering and deposition (Abanyie et al. 2023) in this area due to the geology of the area. So, removing fluoride from groundwater has been identified as a curtail step in the treatment process considering the health risk of excessive fluoride presence in drinking water.

Drinking hard water with a high fluoride content has become one of the major hypotheses for the high incidences of CKDu in Sri Lanka (Indika et al. 2022). According to Figure 4(c) and 4(d), the groundwater in the studied area is contaminated with total hardness (as CaCO₃), and fluoride, which might cause serious health issues if consumed long-term. Therefore, it is necessary to focus on introducing treatment technology to selectively remove problematic ions from groundwater to provide potable water. Therefore, an attempted removal of total hardness (CaCO₃) and fluoride from groundwater was undertaken to ensure drinking water safety for the people in the area.

The influence of voltage on hardness removal performance was tested at 45, 65, and 95 V for three different flow rates (from 8 to 16 m³/h). Figure 5 shows that as voltage increases, the hardness removal also increases. The lowest tested voltage of 45 V at 8 m³/h removed 68.75% by reducing hardness to 110 mg/L, below the SLS 614:2013 limit (250 mg/L). Hardness removal efficiency rises by 77.56 and 86.08% at the same flow rate (8 m³/h) when the applied voltage is 65 or 95 V (Figure 5(a)). Other flow rates (12 and 16 m³/h) show a similar trend of hardness reduction increasing with applied voltage. Doubling the voltage (from 45 V) increases hardness removal efficiency by 17.33%. These results verified that voltage positively affects hardness elimination, as indicated in laboratory-scale research conducted by Karabacakoğlu et al. (2015) and response surface methodology conducted by Bachiri et al. (2024). These results confirm that TED's hardness removal can be improved by increasing the voltage in a similar manner to laboratory-scale studies.

The pattern of hardness removal with respect to flow rate was also studied at three different flow rates: 8, 12, and 16 m³/h (Figure 6). The hardness removal trend shown in the TED process for the flow rate is opposite to the pattern shown with respect to voltage increment. When the flow rate increases from 8 to 16 m³/h, the removal rate of total hardness decreases from 86.08 to 74.72% at the voltage of 95 V (Figure 6(c)). Similar results were observed at other applied voltages. A similar trend of reducing removal efficiencies with increased flow rates was observed by Karimi & Ghassemi (2016). However, this study focused on the individual behavior of calcium and magnesium ions which are the main contributors to hardness in a solution, instead of hardness itself. The increased flow rate reduces hardness removal efficiency due to lower resident time for ions migration to the concentrated chamber at increased flow rates, which has a negative effect on ion removal (Karimi & Ghassemi 2016). Even though a reduction in removal efficiency was revealed with increasing flowrate, all the final concentrations were less than the standard limits of the SLS 614:2013 guidelines (Figure 6), which provided the researchers an opportunity to select the most appropriate flowrate to match the total water demand of the village.

The removal rate of fluoride with increasing applied voltage was studied at three different voltages of 45, 65, and 95 V while keeping a constant flow rate (for three flow rates separately). At a steady flow rate of 8 m³/h, the fluoride removal rate increased from 66.12 to 91.75% when the applied voltage increased from 45 to 95 V (Figure 7(a)). Increasing voltage by roughly twice increased removal efficiency by about 25%. A similar trend of enhancing fluoride removal efficiencies was observed in other studied flow rates as well. This similar pattern of enhancing removal efficiency with increased voltage has been observed in other locations with groundwater, including Turkey (Amor et al. 1998), and Moroccan (Ergun et al. 2008). However, all these studies were laboratory-scale experiments that did not provide real-world applications. Nevertheless, these experiments revealed the importance of selecting an intermediate voltage for long-term performance to maintain process control of the system (Arahman et al. 2016). Thus, raising the voltage excessively may not be beneficial for the long-term running of the ED treatment plant, even though it may improve fluoride removal due to its direct contribution to the energy cost (Tekinalp et al. 2023).

Then, fluoride removal was examined at different flow rates of 8, 12, and 16 m³/h in the diluted chamber. As the flow rate increases, fluoride removal decreases (Figure 8). At 45 V, fluoride removal was 66.12, 55.37, and 45.45% for the flowrates of 8, 12, and 16 m³/h, respectively (Figure 8(a)). Similarly, the removal efficiencies reduced from 82.64 to 59.50% (applied voltage is 65 V, Figure 8(b))) and 91.74 to 70.25% (applied voltage is 95 V, Figure 8(c)) in the other two operated voltages. A similar trend has been reported in the Bambuí aquifer, Brazil water treatment conducted by Patrocínio et al. (2019) using a laboratory-scale ED system (Patrocínio et al. 2019). All these results reveal the importance of maintaining a sufficient residence time to remove an adequate amount of fluoride from the solution during the ED treatment process. Every flowrate tested during this study produces water with sufficient fluoride removal which does not surpass SLS 614:2013 limits (1 mg/L). Therefore, the setup's operating state can be scrutinized by examining additional elements that may directly affect water production costs (Ankoliya et al. 2021).

The concentration of problematic ions (hardness and fluoride) in raw water changes throughout the studied period, as depicted in Figure 4. Therefore, it may not be possible to decide by considering only a single operational condition while the TED runs for the long term. Hence, two different operational conditions were selected to prevent the influence of this seasonal variation on long-term operations. The flow rate was held constant at 12 m³/h, considering the water demand in the village and the yield of the two boreholes. Hence, the voltage was adjusted to achieve different produced water qualities. The intermediate voltages of 45 and 65 V were selected by considering the water quality as well as keeping the TED process control at a manageable range. The qualities of the produced water in the two selected conditions during the initial commissioning period of the TED are summarized in Table 2.

Table 2

Water quality characteristics of raw and treated water produced by TED

Water quality indicator .	Raw water .	Produced water .
		(45 V/7 A) .	(65 V/9A) .
Color platinum-cobalt units (PCU)	0–20	10–12	7–9
Turbidity (NTU)	0.01–11.3	1.21–1.37	0.63–0.75
pH Value	6.43–7.76	8.41	8.52
EC (μs/cm)	600–868	323–441	270–305
Total alkalinity (as CaCO3) (mg/L)	168–390	170–188	130–150
Total hardness (as CaCO3) (mg/L)	150–344	160–180	100–130
Fluoride (as F−) (mg/L)	0.64–1.52	0.66–0.76	0.56–0.68
Chloride (as Cl−) (mg/L)	0–116	18–30	14–20
Sulfate (as ⁠) (mg/L)	0–19	3–6	2–5
Total phosphates (as ⁠) (mg/L)	0.27–1.78	0.3–0.5	0.3–0.4
Iron (as Fe) (mg/L)	0–0.7	0.01–0.04	0.01–0.04
Nitrate (as ⁠) (mg/L)	0.2–7.48	0.2–0.5	0.2–0.5
Nitrite (as ⁠) (mg/L)	0.011–0.073	0–0.1	0–0.1

Water quality indicator .	Raw water .	Produced water .
		(45 V/7 A) .	(65 V/9A) .
Color platinum-cobalt units (PCU)	0–20	10–12	7–9
Turbidity (NTU)	0.01–11.3	1.21–1.37	0.63–0.75
pH Value	6.43–7.76	8.41	8.52
EC (μs/cm)	600–868	323–441	270–305
Total alkalinity (as CaCO3) (mg/L)	168–390	170–188	130–150
Total hardness (as CaCO3) (mg/L)	150–344	160–180	100–130
Fluoride (as F−) (mg/L)	0.64–1.52	0.66–0.76	0.56–0.68
Chloride (as Cl−) (mg/L)	0–116	18–30	14–20
Sulfate (as ⁠) (mg/L)	0–19	3–6	2–5
Total phosphates (as ⁠) (mg/L)	0.27–1.78	0.3–0.5	0.3–0.4
Iron (as Fe) (mg/L)	0–0.7	0.01–0.04	0.01–0.04
Nitrate (as ⁠) (mg/L)	0.2–7.48	0.2–0.5	0.2–0.5
Nitrite (as ⁠) (mg/L)	0.011–0.073	0–0.1	0–0.1

View Large

The higher voltage (65 V) selected water quality was slightly better than the lower voltage (45 V) water quality, as depicted in Table 2, which means that the different produced water qualities can be adapted according to the seasonal variation of groundwater quality in the area. Even though the applied voltage is different, TED provides safe water by effectively removing EC, total alkalinity (as CaCO₃), total hardness (as CaCO₃), and fluoride.

The energy consumption for the two selected operational conditions was calculated according to Equation (3). The energy consumption of TED at the flow rate of 12 m³/h and voltage of 45 V, was 0.00265 kWh/m³. In the other selected operational condition, the energy consumption was 0.00488 kWh/m³ with a flow rate of 12 m³/h and a voltage of 65 V. The calculated operating costs for the two operational conditions were Rs. 0.09/m³ and Rs. 0.17/m³ for 45 and 65 V, respectively. By using less energy (0.00265 kWh/m³) and paying a lower operating cost (Rs. 0.09/m³) under the 45 V operating conditions, a higher water production rate of 90% can be achieved. Hence, operating the system under the operational conditions of 12 m³/h flow rate and 45 V voltage was selected for long-term application.

TED performance was continuously monitored for nearly 3 years under selected operational conditions. The EC of the produced water was less than 500 μs/cm throughout the studied period, confirming the treatment of groundwater through the introduced TED setup. All other water quality parameters are also well below the drinking water standard of SLS 614:2013, including targeted problematic ions such as total hardness, and fluoride. This confirms that the introduced TED provided high-quality drinking water for the villagers by ensuring the safety of drinking water in the area.

A 5-year water quality examination is being conducted using samples from two boreholes in Kahatagasdigiliya. The results show that several water quality parameters exceeded SLS 614:2013 standards, especially hardness (BHI-252.4 mg/L and BH2-235.92 mg/L), fluoride (BHI-1.09 mg/L and BH2-1.01 mg/L), EC (BHI-725.71 μS/cm and BH2- 684.10 μS/cm), and alkalinity (BHI-260.75 mg/L and BH2- 239.93 mg/L). Due to limited economic development and technical expertise, a centralized water supply system was implemented. TED is a mature drinking water treatment method used to selectively remove harmful ions from groundwater during desalination. The mechanism of selective hardness and fluoride removal was also examined, and the high voltage and low flow rate of TED improved ion removal selectivity by 33.99 and 50.45%, respectively. A high-water production rate of 90% can be achieved by consuming a lesser energy of 0.00265 kWh/m³ and a lower running cost of Rs. 0.09/m³ under the operational conditions of 45 and 12 m³/h. It routinely produces 260 m³ of quality drinking water at excellent productivity. The system has been providing water to 1,500 residents in two villages. The equipment has operated steadily and efficiently until today. The above results suggest possible guidance for operating ED toward better selective removal of high hardness and fluoride ions from groundwater to produce quality drinking water. Future work should focus on operational cost and membrane fouling to better understand TEDs for long-term operation as a sustainable solution for groundwater treatment in rural areas.

This research was supported by the Program of the China–Sri Lanka Joint Center for Education and Research by the Chinese Academy of Sciences. It was also financially supported by the Belt and Road Master Fellowship programme (218611422020) and the ANSO Scholarship for Young Talents Award (Series No. 2021ANSOP123). The National Water Supply and Drainage Board, Sri Lanka also cooperated in providing field assistance and information.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.