Laterite (9˚35'N and 5˚23'W), sandstone (5˚21'07'N and 3˚56'30'W) and shale (6˚39'15'N and 4˚58'33'W) samples were collected in Côte d’Ivoire. Each type of material was crushed separately, reduced using a Retsch RS 100 crusher, washed, and then dried for 24 hours. The crushed materials were sieved using Saulas sieves (NF.X 11.501) (see Figure 1 ). Fractions with a particle size of 2 mm were retained to pack the column reactors, following the methodology established by Ezzati et al. [17] . Furthermore, 2 mm size fractions of each adsorbent were calcined at 400˚C for laterite and 300˚C for shale and sandstone, corresponding to the optimal calcination temperatures identified by Koua-Koffi et al. [18] [19] for column tests.

The water samples were collected from wells in the village of M’Badon, situated at 5˚19' and 5˚22'N and 3˚55'N and 3˚56'W, near the former rubbish dump at Akouédo in Côte d’Ivoire. Before each sampling, the drums were rinsed three times with the water intended for sampling. In situ measurements of pH and temperature were taken using a multi-parameter device. The collected water samples were then sent to the laboratory, where the total arsenic content was analyzed using an OPTIMA 2100 dual-view spectrometer (ICP-OES 2100 DV). To adjust the levels of arsenic in the water, a standard arsenic (III) solution was added to each sample can, adhering to the dilution protocols to achieve an arsenic concentration ranging from 1000 to 5000 µg/L for the column experiments.

The experimental setup included PVC columns with an internal diameter of 40 mm and a height of 95 cm, along with a feed barrel, a filter collection tank and a drainage system located at the lower base ( Figure 2 ). Each column was filled from bottom to top with 5 cm of gravel (crushed granite) topped with a geotextile, followed by 30 to 85 cm of adsorbent material. The gravel acted as a drainage layer, while the adsorbents functioned as a filtration medium. The purpose of the geotextile was to prevent the loss of fine particles and to block the water outlets. Contaminated water was introduced into the columns at a flow rate of 5 ml/min. The experiment was deemed complete when the arsenic concentration (Ct) at the column outlet reached 90% of the initial concentration (C0).

During raw material testing, we examined how the initial concentration, the height of the materials and the heat treatment of the materials affected the quality of arsenic removal from water. Regarding the effect of the initial concentration, we packed four columns with 30 cm of each type of raw adsorbent material, which were used to treat water with varying initial arsenic concentrations: 1 mg/L, 3 mg/L, 4 mg/L, and 5 mg/L. To investigate the impact of material height, we identified the highest concentration that showed the least reduction during the first treatment trial and evaluated the influence of different material heights on the purification performance of the columns. This involved filling three columns for each material to heights of 30 cm, 40 cm and 85 cm, respectively. Additionally, heat-treated materials were used in the treatment tests. All experiments were carried out at a feed rate of 5 ml/min.

The total adsorbed As quantity or the maximum of column capacity ( $q_{\text{total}}$), in mg, for a given feed concentration and flow rate ( [20] ), is provided in the following (Equation (1)):

$q_{\text{total}}=Q{C}_0\frac{A}{1000}=\frac{F}{1000}{\displaystyle {\int }_{t=0}^{t=t_{\text{total}}}\left(C_{\text{Ads}}\text{d}t\right)}$ (1)

Here, $t_{\text{total}}$ represents the overall flow duration in minutes, Q symbolizes the flow rate in ml/min, A denotes the total area beneath the breakthrough curve, C₀ is the initial concentration of the adsorbate and q_e (mg/g) is the column adsorption capacity. The theoretical total of As adsorbed by the column is determined by the Equation (2)

$m_{\text{total}}=\left(Co\ast Q\ast t_{\text{total}}\right)/1000$ (2)

With knowledge of the total mass of adsorbed ions and the overall sorbent capacity, one can compute the total percentage of As removed using Equation (3)

$\%\text{Asremoved}=\left(q_{\text{total}}/m_{\text{total}}\right)\ast 100\%$ (3)

Statistical analyzes were performed with R version 3.3.2, with a significance threshold established at p < 0.05. The univariate statistical tests used for data analysis included the Shapiro-Wilk test, Kruskal-Wallis test, t-test, Mann-Whitney test, and Wilcoxon test. The Shapiro-Wilk test was used to assess the normality of the data distribution. Comparisons of arsenic concentrations measured in the filtrates from different materials were performed using both the nonparametric Kruskal-Wallis test and the parametric ANOVA test. Additionally, arsenic concentration comparisons between the filtrates of raw and calcined materials were carried out using the nonparametric Mann-Whitney tests.

The effects of varying the inlet concentration of arsenic ions, set at 1, 3, 4, and 5 mg/L, while maintaining a consistent bed height of 40 cm and a flow rate of 5 mL/min, are illustrated by the breakthrough curves presented in Figure 3 . It is evident that an increase in the inlet arsenic ion concentration results in a steeper slope of the breakthrough curve across all materials, leading to a reduced volume of treated liquid before the adsorbent bed reaches saturation. For laterite, the breakthrough curves for concentrations of 3, 4, and 5 mg/L exhibit similar behavior, maintaining this trend up to a treated volume of less than 20,000 mL, which corresponds to 4000 minutes, before reaching saturation. On the contrary, for an initial concentration of 1 mg/L, laterite achieves saturation after a volume four times greater than that of the other concentrations tested. In the case of sandstone, the curve of 1 mg/L allows a larger volume of water to be treated compared to the other three concentrations. For shale, two distinct families of curves emerge. The first family pertains to concentrations of 3, 4, and 5 mg/L, with all curves reaching saturation after treating a volume of 15,000 mL (3000 minutes). The second family, represented by a concentration of 1 mg/L, has a gentle slope of less than 10%. Here, shale saturation occurs at 7735 minutes, corresponding to a treated volume of 30,000 mL. In particular, for the concentrations 3, 4, and 5 mg/L, the residual concentration did not decrease to 0.01 mg/L.

These findings indicate that a higher initial arsenic concentration leads to a greater driving force for mass transfer, causing the adsorbent to reach saturation more quickly. This results in a reduction in both the breakthrough time and the volume of water treated ( [21] - [23] ). On the contrary, a decrease in arsenic concentration to 1 mg/L produced a later breakthrough curve, with treated water volumes being 9000 mL, 2200 mL and 3000 mL for laterite, sandstone, and shale, respectively. These concentrations align with the WHO guideline value for drinking water at around 0.01 mg/L. This phenomenon may be attributed to the availability of active sites on the surface of the adsorbent in relation to the arsenic levels in the solution. In summary, the initial concentration of adsorbate ions has a significant impact on the breakthrough curve and overall column performance, highlighting the importance of carefully selecting the operating concentration to achieve optimal removal efficiency and breakthrough time [24] .

The influence of bed height on column performance is demonstrated by the saturation times or volumes depicted in the breakthrough curves (see Figure 4 ). For laterite, columns with material heights of 30 cm and 40 cm display similar trends, both reaching saturation after processing volumes below 20 L (approximately 4000 min). On the contrary, the column with a height of 85 cm attains saturation after 5369 minutes, corresponding to a treated volume exceeding 30,000 mL (see table).

The behavior of sandstone is similar, with clear initial slopes observed during the treatment process. In particular, the 8cm column allows for a much larger volume of water treatment compared to the other two heights. In the case of shale, columns containing 30 cm and 40 cm of material reached saturation after treating a volume of less than 15,000 mL (3000 min). On the contrary, the column with 85 cm of material reached saturation after treating 20,000 mL (4000 min). However, the concentration of arsenic in the filtrates did not decline to 0.01 mg/L. During the treatment test, significant differences in arsenic concentrations were observed in the filtrates from the various crude laterite columns (Kruskal-Wallis test: p < 0.05). In particular, a significant difference was identified between heights of 30 cm and 85 cm, as well as between 40 cm and 85 cm (Mann-Whitney test: p < 0.05). The Kruskal-Wallis test (p < 0.05) also revealed differences in filtrates from columns packed with raw sandstone, specifically between columns of 30 cm and 85 cm (Mann-Whitney test: p < 0.05). ANOVA analysis showed a significant variation in residual arsenic concentrations between different reactors for filtrates treated with raw shale (p < 0.05), with differences observed between the 30 cm and 85 cm filtrates, as well as between the columns 40 cm and 85 cm (p < 0.05). In general, using bed heights of 30 cm, 40 cm, and 85 cm resulted in an increase in the breakthrough time and a higher percentage of arsenic removal as the volume of water treated increased ( Table 1 ).

This phenomenon can be attributed to the increased surface area and the greater number of binding sites available for adsorption as the height of the bed increases [25] .

The enhancement of the height of the bed for the removal of arsenic leads to a larger number of accessible sites for adsorption, as well as an extended contact time between the arsenic ions and the adsorbent material [26] .

Similar findings were documented by Tan et al. [27] , Han et al. [28] , and Sadaf and Bhatti [29] , who observed comparable results during the adsorption of basic dye using activated carbon derived from oil palm husks, the adsorption of methylene blue using phoenix tree leaf powder in a fixed bed column, and the in-column removal of indosol from groundnut husk, respectively. Roy et al. [30] and Maji et al. [31] also reported similar results when removing arsenic from a column using sugarcane charcoal and natural rock coated with iron oxide, respectively. Higher bed columns demonstrate improved performance, as noted in the literature [32] . Furthermore, the slower exhaustion of the adsorbent bed is advantageous for the effective removal of arsenic ions from water. Therefore, greater bed heights are preferred, as they offer a longer duration of effective adsorption [33] .

q_total = overall arsenic ion adsorption capacity (mg), m_total = overall mass of adsorbed arsenic (mg), V_b = volumetric output at breakthrough point (ml), V_T = total volumetric output at saturation (ml), q_e = adsorption capacity (mg/g).

The quantity of arsenic adsorbed and the volume of water treated with the calcined adsorbent within the columns are shown in Figure 5 . For columns containing 85 cm of material, the arsenic concentration at the column outlet remained below 0.01 mg/L for treated water volumes of 7460 mL for laterite, 3510 mL for sandstone, and 4400 mL for shale. The calcined materials demonstrated a lower residual arsenic concentration compared to the raw materials. The effectiveness of the materials was classified as follows: sandstone < shale < laterite (see Table 1 ). Statistical analysis revealed a significant difference between columns filled with raw materials and those with calcined materials (Kruskal-Wallis test: p < 0.05). This difference was notably significant in the filtrates of the 85 cm columns containing raw materials compared to those containing calcined materials (Mann-Whitney test: p < 0.05).

The favorable results obtained with calcined materials can be attributed to the alteration, disappearance, or modification of certain minerals due to the effects of temperature [34] , as well as the presence of minerals such as hematite, kaolinite, gibbsite, calcite, and goethite [18] , [19] . Valix and Cheung [35] noted that calcination can influence both the chemical components and the physical structure of the adsorbent. This process effectively removes organic matter while altering the structure of the adsorbent, thereby enhancing its capacity for arsenic adsorption [36] . These adsorption values are lower compared to those achieved with granular ferric hydroxyl (GFH), which is a certified product for the removal of arsenic in water [37] .