The study was conducted in the Kandé and Lyndiane neighbourhoods (commune of Ziguinchor, Lower Casamance, Senegal). Samples were collected over a four-month period, covering both the rainy season and the dry season, from four wells: two per neighbourhood, one located in the centre and the other on the periphery. After each sampling, the physico-chemical parameters electrical conductivity (EC), pH, total dissolved solids (TDS), temperature (T) and a complementary measurement unit (U) were measured in situ using a multiparameter conductivity meter (Figure 1).

Figure 1. Multiparameter conductivity meter.

Ditakh shells (Figure 2(a)) were collected, crushed and carefully wrapped in aluminium foil before being placed in a muffle furnace (Figure 2(b)) for the carbonisation step. The carbon obtained from carbonisation then underwent chemical activation: one portion was treated with lemon juice (Figure 2(c)) and the other with phosphoric acid (H₃PO₄) (Figure 2(d)). The resulting carbons were washed repeatedly until the wash filtrate reached a neutral pH, then oven-dried at 105˚C for 24 h. All quantities of material used in the experiments were weighed using an electronic balance.

Figure 2. Equipment and chemical reagents used for sample conversion.

Two complementary approaches were used to evaluate the performance of the activated carbon:

Direct contact: the activated char was brought into contact with the water to be treated, with parameters (pH, EC, TDS, T, U) measured every 30 minutes, in order to determine the volume of water that a given mass of char could treat and to identify the optimal mass (25, 30 and 35 g for Kandé; 2, 3 and 4 g for Lyndiane), as well as the effect of the activating chemical (lemon juice vs. H₃PO₄);Multilayer filtration: water was percolated through a filter bed of superposed sand, limestone and activated char, under two configurations Entity 1 (2 layers of sand + 1 layer of char + 1 layer of limestone) and Entity 2 (2 layers of char + 1 layer of sand + 1 layer of limestone) with the outlet filtrate parameters measured (Figure 3).

Figure 3. Filter components sand (a), limestone (b), activated char (c) and the superposed filter bed (d).

In this study, the results for pH and EC will be presented because TDS (mg/l) and U (electrical potential in millivolts, mV) are parameters related to ion variation (TDS is proportional to EC). However, the temperatures obtained range from 25˚C to 27.4˚C, so they remain within the WHO recommended range. To make the document less bulky, we focused on pH and EC, which provide more physicochemical insight into the samples.

All results were compared against WHO potability standards, in particular the electrical-conductivity limit of 1300 µS/cm.

Monitoring of pH over four months shows that Kandé well water is generally acidic, with neutrality observed only in September and October, the rainy period (Figure 4). This transient neutralisation could be explained by rainwater intrusion into the well, diluting H₃O⁺ ions or introducing cations. In Lyndiane, the same seasonal pattern is observed for pH, with neutrality during the rainy season (September-October) followed by a return to acidity in the dry season (Figure 4(b)).

Figure 4. pH Variation of Kandé and Lyndiane well water, (a) and (b) respectively.

Monitoring of the electrical conductivity of Kandé water reveals values exceeding the WHO guideline (1300 µS/cm) (WHO, 2017), reflecting strong ineralized on of this water (Figure 5(a)).

The electrical conductivity of Lyndiane water generally complies with the WHO guideline, except for the peripheral well, whose water is strongly ineralized (Figure 5(b)).

Figure 5. EC variation of Kandé and Lyndiane well water, (a) and (b) respectively.

These findings are consistent with those reported in other West African contexts, where the acidity and mineralisation of well water are frequently associated with the nature of the geological substrate and the seasonal dynamics of the water table [2][4].

The spatio-temporal analysis of electrical conductivity (EC) in Kandé and Lyndiane shows that the town centre, characterised by higher population density, systematically displays higher conductivity values than the periphery (ECc > ECp) throughout the study period (September to December). In African urban and peri-urban hydrogeology, this spatial differentiation is explained by the interaction of anthropogenic and climatic factors. This anthropogenic-enrichment phenomenon drastically increases the overall ionic load of the water and hence its electrical conductivity. Conversely, although the dense urban centre is subject to strong anthropogenic pressure linked to on-site sanitation systems (soak-away latrines), the pollution there tends to be more localised and mainly composed of organic or bacteriological matter, together with an increase in total mineralisation.

Overall, in light of the interpretation criteria for physico-chemical water parameters ([9][10]), these results confirm the need to treat Kandé and Lyndiane water before any consumption.

These observations are consistent with recent regional hydrogeochemical investigations, which attribute the mineralisation of shallow groundwater in Casamance to a combination of silicate weathering, carbonate dissolution and cation-exchange reactions, modulated by seasonal recharge and, in the most exposed coastal aquifers, by saltwater intrusion [11]. The transient neutralisation observed here during the rainy season is compatible with a dilution effect from meteoric recharge diluting acidic and saline end-members, a mechanism also documented for other mineralised aquifers of the Senegalese coastal zone [12]. It should be noted, however, that without major-ion or isotopic data the present study cannot formally distinguish between dilution by rainwater and a possible seasonal contribution of saline intrusion, and this remains a limitation of the physico-chemical monitoring performed here.

For Kandé water treated with 30 g of lemon-juice-activated char, the evolution of pH as a function of the volume of water treated shows two phases (Figure 6(a)): an initial increase, followed by a slight decrease beyond 290 mL. The increase would be related to the release, by the char, of internal basic ash (calcium, potassium and magnesium carbonates), which raises the basicity of the solution. The decrease observed beyond 290 mL would reflect the carbon reaching its adsorption limit, this volume thus representing the treatment capacity of 30 g of activated char for this water a saturation behaviour already reported for other plant-derived activated chars [8][9]. At the same time, electrical conductivity decreases progressively with the volume treated (Figure 6(b)), consistent with effective adsorption of dissolved ionic species.

Figure 6. Variation of pH and EC of Kandé water after treatment with activated char (a and b respectively).

For Lyndiane water, electrical conductivity also decreases at first, before rising slightly beyond 290 mL (Figure 7(b)), which could be explained by partial release of previously adsorbed pollutants once the char becomes saturated a classic desorption phenomenon once the char’s maximum capacity is reached. This volume of 290 mL therefore constitutes the treatable volume for 3 g of lemon-juice-activated char for this water (Figure 7(a)). These results confirm that knowledge of the water-volume/char-mass pairing is decisive in ensuring treatment effectiveness.

Figure 7. pH and EC of Lyndiane water after treatment with activated char (a and b respectively).

Analysis of Figure 8 helps us to understand the impact of the mass of material (25 g, 30 g, 35 g) and the volume of treated Kandé water on two key physico-chemical parameters: pH and electrical conductivity (EC). The treatment causes the water to become progressively more alkaline. This effect is directly proportional to the quantity of material used.

Figure 8. Effect of AC quantity on the pH of Kandé water.

The EC decreases markedly (Figure 9) for all masses (a drop of 58 to 108 μS/cm). This sharp decrease in the concentration of free ions in the water indicates a process of chemical precipitation or adsorption onto the material. The curve for 35 g consistently remains above that for 25 g. This shows that the increase in mass initially introduces more soluble ions into the medium, even though the downward trend remains the same. From 320 mL onwards, the EC stops decreasing and begins to rise slightly. This behaviour reflects the saturation of the binding sites on the activated char, or the depletion of the reagent limiting the precipitation.

Figure 9. Effect of AC quantity on the EC of Kandé water.

For Lyndiane water treated with H₃PO₄-activated char, the mass of 4 g, among 2, 3, and 4 g, optimized both pH (Figure 10) and conductivity (Figure 11). Therefore, for each water type, there is an optimal mass of char beyond which treatment efficiency no longer improves significantly. This is a typical behavior of adsorption phenomena before saturation of active sites [8][9]. As the treated water volume increased, pH increased (Figure 10), while electrical conductivity decreased (Figure 11). This inverse relationship suggests that chemical reactions release hydroxyl ions (OH⁻) or consume hydrogen ions (H⁺), while causing the precipitation or retention of certain dissolved salts, thereby reducing the overall EC of the water.

For the same water volume, a higher mass, 4 g, resulted in a higher pH and higher EC than a lower mass, 2 g. Increasing the mass provides a larger contact surface and more active sites. The increase in pH observed for both water samples is consistent with the general mechanism described for carbon-rich materials obtained by pyrolysis: their ash fraction contains carbonates and soluble oxides of calcium, potassium and magnesium, which neutralise H⁺ ions and raise the pH of the solution with which they come into contact, an effect whose magnitude has been found to correlate more closely with the alkalinity of the carbonised material than with its own intrinsic pH ([13][14]).

Figure 10. Effect of AC quantity on the pH of Lyndiane water.

Figure 11. Effect of AC quantity on the EC of Lyndiane water.

This accelerates the release of alkaline compounds, increasing pH, and initially increases the quantity of ions released into the solution, increasing EC. The pH continued to rise and EC continued to decrease even at 350 mL, without reaching a perfect plateau, although the slope of the evolution varied depending on the mass. Evaluation of the saturation limit of the material may be conducted by testing much larger water volumes, for example greater than 500 mL, in order to identify the point at which treatment loses efficiency through stabilization of pH and EC.

Comparison of the two activation modes shows that H₃PO₄-activated char continues to treat the water beyond the point at which lemon-juice-activated char reaches saturation (around 370 mL for Kandé water), for both pH and electrical conductivity (Figure 12 and Figure 13).

Figure 12. Effect of activation type on the pH of Kandé water.

This behaviour could be explained by a more developed porosity in the phosphoric-acid-activated char, giving it a higher adsorption capacity [6][7].

Figure 13. Effect of activation type on the EC of Kandé water.

The same observation applies to Lyndiane water, where H₃PO₄-activated char gives the pH values closest to neutrality (Figure 14) and the variation of the EC is shown in Figure 15. This result is consistent with the literature, which generally associates phosphoric-acid activation with the development of microporosity and a high specific surface area, favourable to the adsorption of organic and mineral pollutants [6][7].

Figure 14. Effect of activation type on the pH of Lyndiane water.

Figure 15. Effect of activation type on the EC of Lyndiane water.

This observation is in line with the broader literature on chemically activated char: phosphoric acid is widely regarded as one of the most effective agents for developing microporosity and mesoporosity in lignocellulosic precursors, with specific surface areas reported across a wide range of roughly 450 to 2800 m²/g depending on the precursor and the acid-to-precursor impregnation ratio, generally exceeding what is achieved with weaker organic-acid activation [15]. Lemon juice, whose active agent is citric acid, is a much weaker acid than H₃PO₄ and predominantly promotes surface oxidation rather than the extensive pore-opening associated with phosphoric-acid impregnation, which is consistent with the more limited and more rapidly saturated performance of the lemon-juice-activated char observed in this study.

The multilayer filtration setup confirms the benefit of combining activated char with other filter media. For both neighbourhoods, the configuration with two layers of sand (Entity 1) gives pH values closest to neutrality (Figure 16), Figure 17 illustrates the influence of the filter bed configuration (sand versus carbon) on the change in the pH of the Lyndiane water as a function of the volume filtered. The results show a gradual increase in pH for both systems as the volume of water increases from 110 to 350 mL. This continuous alkalisation of the effluent suggests a gradual release of basic species or an ion exchange process within the filter media as the water passes through. The configuration comprising two beds of activated carbon (Entity 2) consistently produces a higher pH than the sand-based system (Entity 1), with final values reaching 6.89 and 6.75 respectively. This superior performance of activated carbon can be explained by its naturally basic surface chemistry (the presence of surface functional groups such as hydroxyl or pyrone groups), which is capable of trapping H⁺ ions or releasing OH⁻ ions. Although both configurations tend to neutralise the initial acidity of the Lyndiane water to bring it closer to drinking water standards (pH between 6.5 and 8.5), the carbon bed proves more effective from the very first volumes filtered, offering greater buffering and neutralising capacity than the sand. And the lowest electrical conductivities (Figure 18, Figure 19), compared with the configuration containing two layers of char (Entity 2).

From a comparative perspective, these results are consistent with the work in [16] and with studies on filtration using mixed granular media, which highlight that the addition or exclusive use of activated char alters the carbonate/bicarbonate equilibrium of the treated water more significantly than simple physical filtration using sand.

This superior performance of sand could be explained by its fine grain size and the low percolation rate it imposes, favouring a longer contact time and hence better pollutant retention [9][10]. Conversely, the coarser grain size of the char in Entity 2 accelerates water passage, which reduces contact time and limits pollutant retention.

This finding is consistent with the mechanisms described for household slow sand filters, in which the fine grain size of sand and the correspondingly slow percolation rate promote extended contact time between water and the filter medium a condition favourable both to physicochemical retention and to the development of a biologically active surface layer capable of retaining dissolved and particulate contaminants [17]. Increasing the proportion of coarser-grained carbon, as in Entity 2, shortens the residence time of water within the bed and therefore curtails the diffusion-limited processes that adsorption efficiency depends on.

Taken together, the results show that activated char produced from ditakh shells, regardless of the activation mode, substantially improves the physico-chemical quality of Kandé and Lyndiane well water. Three factors determine treatment effectiveness: the mass of char used relative to the volume of water to be treated; the type of chemical activation, phosphoric acid conferring greater porosity and hence higher adsorption capacity than lemon juice; and the configuration of the filtration setup, with a sufficient quantity of sand improving pollutant retention, three levers already identified as decisive in optimising plant-derived activated chars [8][9].

Figure 16. Effect of bed configuration on the pH of Kandé water.

Figure 17. Effect of bed configuration on the pH of Lyndiane water.

Figure 18. Effect of bed configuration on the EC of Kandé water.

Figure 19. Effect of bed configuration on the EC of Lyndiane water.