Iron is present in some sites in drinking water intended for the consumption of the Senegalese population, at concentrations that exceed WHO water quality standards (0.3 mg/L) (Ahmed and Mir, 2022). Dissolved oxygen precipitates Fe(II) to Fe(III) and imparts a reddish color, metallic taste and undesirable odor to the water, domestic softeners become clogged with Fe (III) precipitates, the formation of precipitates in the pipes reduces the inside diameter and increases energy losses. When the ferrobacteria formed in the pipes die and disappear, bad odors and unpleasant tastes can be caused. Iron oxidation rates increase with the amount of oxygen present in the medium, pH and redox potential can reach conversion rates of around 90% at pH=7 and Redox potential = 400 mV (Elliott et al., 2014). Oxidation of iron is markedly slower at pH<6 and precipitates may persist for some time in aerated waters. Oxidation yields can be increased by the presence and action of certain microorganisms (Bray et al., 2017; Nayyeri et al., 2021). There are several kinds of bacteria that oxidize dissolved Fe (II) by different mechanisms. Indeed, microorganisms in particular Leptothrix ochracea, Crenothrix polyspora and Gallionella species can cause biological oxidation if the operating conditions are established.

The aeration operation followed by solid-liquid separation on a sand bed is the most widely used chemical treatment method in groundwater treatment plants in Senegal. Michalakos et al. (1997)have shown in their scientific work that aeration is the adequate method for the oxidation of iron in groundwater where concentrations are greater than 5 mg/L to avoid chemical costs and the complexity of technology. Separation from liquid to solid phase is possible using sand filters. If necessary, other treatments can be added, such as chemical oxidation with chlorine (Vayenas and Lyberatos, 2005), potassium permanganate or ozone, ion exchange processes and biological oxidation (Cai et al., 2015).

In this work, the treatment method that relied on a continuously operated fixed bed column to remove dissolved Fe (II) was used. The column was filled with plastic coated jujube seeds with an average diameter of 8.08 mm, used as the filter material. The limits between the physical, chemical and biological removal of iron are not identified in this work. Nevertheless, the conditions in the field of biological oxidation are fixed. Experimental tests are carried out on the determination of the bed backwashing operating cycles, the oxidation rates of Fe (II), the effect of the Fe (II) concentration, the influence of the height of the filtration bed and the characterization of the withdrawn sludge, resulting from the biological treatment of groundwater.

The biological filter used at the pilot scale is a non-transparent PVC tube with an internal diameter of 40 mm and 90 cm in height with an ascending type supply. The height of the pilot column is half that of full-scale industrial filters measuring on average 1.80 m (Katsoyiannis and Zouboulis, 2004). Along the depth of the filter there are 3 orifice ports for the determination of the iron content which passes through the bacterial bed. The aeration takes place at the start of the process by a MECAFER type air compressor at a constant flow rate of 760 L/h (Figure 1). Once the pseudo-continuous speed is reached, the air supply to the compressor is cut off. The natural air draft mode is used and no external source of mechanical ventilation is used in the biofilter. The advantage of using this type of biofilter is that it does not always require an external air supply, because air enters naturally through the filter due to the temperature gradient between the inside and the outside of the biofilter.

A hydraulic head of 261.67 m³/m² per day was used in the experimental work to avoid flooding the biofilter. Water doped with iron to filter through biological filter body, could flow in the same direction as compressed air. Backwashing of the filters was necessary due to the clogging of the pores due to the growth of biomass and the deposition of iron precipitates on the surfaces of the coated jujube seeds. The backwashing was performed, according to operating conditions, using high flow velocities of air flow. Before to sampling, the biofilter was backwashed with a solution of 1 mg/L concentration and then operated in continuous mode while maintaining the feed conditions for at least 24 h, to ensure pseudo-stable conditions, to stabilize the distribution of the biofilm population as much as possible (Tekerlekopoulou et al., 2008). A syringe was immersed up to the center of the filter for the collection of water samples. The biological filter is filled with coated jujube seeds with an average diameter of 8.08 mm, a specific surface area (A_s) of 548.13 cm²/cm³ and a porosity of 0.423. On this support material, microorganisms responsible for removing iron from the water to be treated are cultivated (fixed culture), and were used for several days (2 months). In all the experiments, the water temperature is practically constant (25± 1°C). The oxygen dissolved in the liquid over the entire depth of the filter was between 6 and 8 mg/L. We have also tried to vary the redox potential from 200 to 400 mV, to be sure of being in the domain of biological iron removal. Once the shake flask culture reached the stationary phase (measured by a wavelength of λ_max = 530 nm), the suspension was transferred to the bioreactor. The volume of cell suspension added to the bioreactor was 1 L (Zhang et al., 2002).

To start the experiments, a start-up time of 10 days was necessary for the maturation of the biofilter. As earlier stated in batch mode after 10 days, microscopic observations were made. During continuous operation, solutions were added at iron concentrations between 0.5 and 2 mg/L for the column feed. The feed solution was the result of mixing a solution of ferrous iron prepared from iron sulfate (containing small amounts of ammonia and phosphorus) (Tekerlekopoulou et al., 2006).

For the laboratory experiments of this study based on monitoring the kinetics of biological degradation of iron, the measurement technique by UV-Visible spectrophotometry was used, in particular the ortho-phenanthroline method. The total microflora count in the sludge sample was made on the nutrient agar medium (GN), the number of germs in our sample was determined according to the following rule:

Where N: number of microorganisms in CFU/mL; n: number of colonies counted; V: volume sampled (0.1 mL); d: dilution.

For the tests, we chose to work under operating conditions with a volumetric flow rate of the sample 8.11 L/h and an air supply flow rate Q_G = 760 L/h to avoid entraining the ferrobacteria cultivated and to have a good residence time. However, different flow rates will be tested in order to gauge the efficiency according to the feed flow rates (Sample-Air). This is because the pilot scale ascending filter was kept in continuous operation for two and a half months to ensure its stable state and the development of the biofilm on the outer walls of the coated jujube seeds.

In all of the experiments, the water temperature was almost constant (25±1°C), as was the ambient temperature in the laboratory. The pH in the liquid phase over the entire depth of the filter is between 9.5 and 10. The redox potential considered is 300 mV in this study with a dissolved oxygen concentration between 6 and 7.5 mg/L. Under these operating conditions, and since the measured redox potential varies from 0.2 to 0.4 V, the oxidation of biological iron is predominant.

Determination of bed backwashing operating cycles

Once the seeding is done and the continuous mode established, the inlet flow rate of the water to be treated is set at 8.11 L/h. For iron concentrations of 0.3 to 0.6 mg/L, no backwashing was necessary (Figure 2). It appears that a state of equilibrium has been established in the biofilter. For higher iron concentrations, the duration of the operating cycles becomes short. For 2 mg/L, an operating cycle time of one day was found to be sufficient, while the iron concentration at the top of the column was approximately 0.983 mg/L (limited operation). These experiments verify the work of Michalakos et al. (1997) which showed that generally, an increase iron concentration decreases the required duration of the operating cycles. At the exit of the biofilter, suspended solids or iron precipitates of small sizes could be observed, while the end of each cycle is related to a decrease in outlet volume flow rate at the top of the column. This was used as a criterion to start the backwash of the biological filter.

Iron concentration profiles in the biofilter

The biologically active filter for iron removal, previously used in experiments for determining bed backwash duty cycles, was used to assess iron concentration profiles in the biofilter at different dates after backwashing of the bed filtration. Monitoring the pilot column will allow us to study the iron removal efficiency over the 24 h period of continuous operation. The effectiveness of the biofilter is illustrated in Figure 3 for a theoretical initial concentration of 2 mg/L. At the start of a filtration cycle (5 h after backwashing), the iron concentration decreases but well below the limit set by WHO for drinkability recommendations. This profile is practically constant after 15 and 20 h of filter operation, thus showing a tendency towards saturation of the biofilter.

Iron concentration profiles in the filter at different air flow rates

Monitoring of the filtration cycle (total concentration of iron at the outlet of the 3 intake points of the column) at different air supply rates (760, 900 and 1200 L/h) for a supply rate of l the constant sample and equal to Q_L=8.11 L/h a shown, even at high air flow rate, the iron concentration in the collected sample decreases with the height of the bed (Figure 4). This has decreased and is slightly approaching the standard value recommended by the WHO at a filter height of 80 cm for a residence time of 24 h and a flow rate Q_G = 760 L/h. A slight shift in concentration profiles is observed with increasing air supply flow rates (900 and 1200 L/h). However, for an air flow greater than 1200 L/h the pilot column is unstable with the abundant presence of precipitates formed in the sample collected at the top of the column. These results are similar to those of Benjwal and Kar (2015) and Benjwal et al. (2016)which were obtained following the study of the iron concentration profiles in the biofilter they designed. Thus, since the feed rate supplied by the air compressor Q_G =760 L/h seems to be more adequate for the proper functioning of our pilot column, the biological treatment of iron was carried out at this rate.

According to the work of Štembal et al. (2005), continuous air aeration is important for the correct operation of a co-current biofilter. Tekerlekopoulou and Vayenas (2007)showed that the main factor responsible for airflow in an open-top biological filter is natural draft. The driving force behind the air flow is caused by the temperature difference between the ambient air and the air inside the pores of the filter bed. Thus, the use of an ascending filter has the advantage of not requiring an external air supply or a continuous ventilation system (air compressor).

Iron concentration profiles in the filter at different sample flow rates

Figure 5 shows the evolution of the iron content in the column at different volumetric flow rates. Unlike studies of iron concentration profiles in the biofilter at different air flow rates, we noticed, for a constant air supply rate Q_G=760 L/h, that the increase in the supply rate of the sample decreases iron removal rates. This is because increasing the sample flow rate decreases the residence time in the biological bed. This study confirms that Q_L = 8.11 L/h is more adequate for proper operation of the pilot column.

The experimental data of the current study are in very satisfied agreement with the prediction of the model that was developed in the studies of Štembal et al. (2005). Their research showed that an increase in volumetric flow reduces the efficiency of the filter. In particular, for a feed iron concentration of 2 mg/L, an increase in flow from 1000 to 2000 mL. min^-1 reduces filter efficiency from 100 to 81% (Tekerlekopoulou et al., 2006).

Biological oxidation kinetics of iron

It is established that iron can be removed from groundwater through the application of biological processes (Katsoyiannis and Zouboulis, 2004). It is also documented that iron oxidation rates are much higher in the process of chemical removal. Iron removal was compared to values reported in the literature in this study. These experiments were performed by adding Fe (II) to distilled water, at different initial iron concentrations (Figures 6 and 7).

According to the experimental data (Figure 6), the effect of varying the initial iron concentration improves the efficiency of the biofilter by approximately 5.84 to 4.47% under the particular experimental conditions and concentration range of 1 to 1.5 mg/L and 1.5 to 2 mg/L. The efficiency of the biofilter increased by 5% on average. However, increasing the depth of our biofilter would be necessary to reach the authorized limit in drinking water. Tests carried out by Cai et al. (2015) and Tekerlekopoulou and Vayenas (2008)show that a decrease in the entry concentration of 1.5 to 1 mg/L, increases the effect of biological oxidation or the efficiency of the biofilter by about 5% while reducing the required depth by about 40%. The rate of oxidation is a function of the content of ferrous ions but independent of the amount of ferro-oxidant bacteria in the environment (Winklehaus et al., 1966).

Correlation of the curves revealed that the biological oxidation of iron in the pilot column also followed a first order kinetic rate, given by the following equation:

For a constant pH and iron concentration in the sample, the kinetic constants are equal to 7.52, 10.86 and 13.054 per day, respectively for the initial theoretical iron concentrations [Fe²⁺]₀ =1, 1.5 and 2 mg/L. This clearly shows that the biological oxidation of iron is a slow reaction, under the set experimental conditions, occurring in the reconstituted water to be treated. This can also be indicated by the calculated half-life constant for iron oxidation, which was found to be 21, 28 and 34 days.

Compared to the work of Katsoyiannis and Zouboulis (2004) and Vaclavikova et al. (2008)the oxidation of iron with L. ochracea and Gallionella ferruginea, was far faster than ours. Biological oxidation of iron at pH=7.2 showed a half-life of 12 min (Gleeson et al., 2012; Van Beek et al., 2012). This could be explained by the catalysis of FeOOH precipitates, which can increase the removal rate by a factor of 10. This has also been observed in studies by Gleeson and Wada (2013), Taylor et al. (2013) and Wada et al. (2010).

Biological iron oxidation kinetics and biomass monitoring

Experiments based on varying the theoretical initial iron concentration of 2 mg/L, was carried out in the pilot column under prolonged aeration and the presence of isolated ferrobacteria (Figure 8). We studied the biological oxidation kinetics of iron and the development of biomass as a function of bed contact time. The liquid samples withdrawn from the top of the column, after filtration on filter paper, were analyzed by the UV-Visible spectrophotometric technique. To determine the concentration of biomass in our experiments, 20 mL samples were taken and then filtered through filter paper to remove the precipitated iron. Liquid samples were filtered again through filter paper to estimate the bacterial

dry weight on the volume collected.

Figure 8 shows the iron concentration profiles as a function of the contact time with the bed, in particular the depletion of iron in the medium, the concentration of biomass and the application of the kinetic model of Tekerlekopoulou and Vayenas (2008). These results indicate that a decrease in concentration leads to an increase in biomass. The increased biomass significantly induces precipitates of iron and biomass content in the samples collected. This could be due to the detachment of bacteria colonies at the bottom of the column caused by the air supply. We also notice a decrease in filtration speed as a function of time. Indeed, the trapped biomass further reduces the empty space between the coated jujube seed particles and the filtration rate thus becomes the dominant parameter to describe the deposition of iron precipitates. The biological oxidation of iron can be expressed by the expression of the kinetics of Tekerlekopoulou and Vayenas (2008):

The kinetic constants obtained are: μ_max = 3.2 L/day, K_S = 0.826.10^-3 mg/cm³, Y_Fe = 16.72 mg Cells/mg of Fe (II) oxidized.

Figure 8 also shows the efficiency of iron removal in the biofilter. This model makes it possible to evaluate a complete oxidation of iron. The data from this model can facilitate the sizing of the pilot columns and the predictions of the effectiveness of biological iron treatments according to Tekerlekopoulou and Vayenas (2008) who used it in their research. In the context of our study, it is evident that the monolayer biofilter is not effective according to the operating conditions tested, because for all the iron contents chosen, we do not manage to reach the standards of drinkability recommended by WHO.

Iron concentration profiles along the depth of the single-layer filter

The effect of the initial concentration of iron along the depth of the monolayer filter was carried out in order to study the performance of the biofilter (Figure 9).

Figure 9 refers to the oxidation of iron for a constant sample volumetric flow rate Q_L =8.11 L/h and theoretical iron concentrations of 1, 1.5 and 2 mg/L. It is evident that an increase in the iron concentration at the entrance of 1 to 2 mg/L increases the depth of the biofilter needed to achieve safe iron removal from 26 to 60 cm. The rate of iron oxidation (mg Fe²⁺/m².day) increases accordingly with increasing concentration. In particular, an increase in iron concentration from 1 to 1.5 mg/L and finally to 2 mg/L increases the rate of iron oxidation from 87.535 to 572.957 and 891.267 mg Fe(II)/m².day, respectively.

Overall, the general observation that has been made shows that at the top of the filter, the efficiency of the filter is not sufficient, which implies that the height of the biofilter bed is a crucial parameter for the biological oxidation of iron. On the wafer from the top half of the filter, the iron oxidation efficiency increases to about 7.54% on average between the mid-column set point and the outlet. A general observation of the redox potential at the exit of the top of the column increased significantly. This is consistent with the remarks of Mouchet (1992)that the redox potential increases due to the oxidation of iron. Indeed, the biological oxidation of iron does not require more stringent conditions than the biological oxidation of manganese and ammonia with a redox potential greater than 300 to 500 mV.

Determining the depth of the filtration bed

For the study of the optimal bed height of our pilot column, we used the concentration profile equation for the biological reaction in a packed-bed bioreactor developed by Štembal et al. (2005):

The experimental values of the term (k/Uⁿ) found by Štembal et al. (2005) are grouped together in Table 1.

The dimensions and optimal operating conditions of the biofilter according to the composition of the groundwater to be treated made it possible to show the importance that binds the kinetic parameters. The kinetics of removal of contaminants during groundwater treatment is considered to be an important issue, as it could provide information on the time required to effectively remove the specific contaminant, which is required to size treatment units (Table 2).

Application to Senegalese groundwater

The main objective of this paper was to achieve the installation of a pilot for the biological treatment of excess iron in groundwater in Senegal. In this wake, after the installation and start-up of the biological reactor in continuous mode, we carried out tests on groundwater sampled at Dakar-Pout Kirene (Drilling PK2), South Dakar-Pout (Drilling PS5) and in Kaolack-Koungheul (Drilling KK5). In addition, in previous tests, iron concentrations were determined from a limited number of analyses of artificial water spiked with iron sulphate (FeSO₄). It therefore seemed advisable to us to undertake a more systematic study, based on water intended for the production of drinking water. In order to study at best a typical water coming from soils rich in iron, while taking into account the different ions present, we tried to make a biological removal of iron from some groundwater which comes to us from different areas mentioned earlier. We started from an original compilation of groundwater compositions intended for the production of drinking water, whose catchments contain an iron concentration between 0.5 and 2 mg/L. This will allow us to better justify the advantages and disadvantages of the realistic application of this method, in the field, for the treatment of groundwater.

Physical and chemical characterization of groundwater

The analysis of groundwater from boreholes PK2, PS5 and KK5 was carried out at the Senegalese water control laboratory. The results of the specific physical and chemical parameters of these waters are summarized in Table 3.

Biological removal of iron from groundwater

The experiences previously developed in continuous operating mode are repeated with groundwater. The treatment was carried out according to the optimized operating parameters and the results obtained are shown in Tables 4 and 5.

The results of continuous treatment show that it is possible to treat reconstituted water. However, the application to this groundwater gives values ??after treatment which are satisfactory but do not comply with the limit of the WHO standard recommended in drinking water. Table 5 shows that the biological treatment gives results of reduction percentages of 39.3, 48.76 and 67.23%, respectively for the PK2, PS5 and KK5 boreholes. The difficulties of treatment for biological removal appear mainly at the level of groundwater with an increasingly important iron concentration. Figure 10 shows the bar graph of groundwater treatment from boreholes PK2, PS5 and KK5 at the inlet and outlet of the biofilter. Indeed, after 14 days of contact time (continuous recirculation in a closed circuit), the greatest removal takes place for the initial concentration of 0.94 mg/L (Table 5). This means that for a high initial oxidation rate, the capacity will not necessarily be the greatest after 14 days of contact time. This confirms the biological treatment tests which were carried out in tubes.

Nevertheless, we record low biological removal percentages. This could be linked to the electrical conductivity, the very high turbidity in these collected groundwater, which exceeds the WHO standard standards (COND: ≤ 300 µs/cm Turb: ≤5 NTU), as well as to the influence metals dissolved in water. This drawback means that we recommend regulating these parameters before proceeding with the biological treatment of iron.

Characterization of the withdrawn sludge: Case of drilling PK2

The characterization of the products of biological oxidation has made it possible to show that with the X-ray fluorescence technique, the oxidized products contain a significant amount of ferric iron Fe (III). These results are in agreement with those of Katsoyiannis and Zouboulis (2004), where it was reported that the oxidation of Fe (II) by oxygen in reconstituted solutions mainly gives Fe(OH)₃ (Figure 11).

The iron peak is observed at 6.305 keV (Figure 12). According to previous work published by Katsoyiannis and Zouboulis (2004), this is the characteristic energy of iron oxides, probably goethite or ferrihydrite. Goethite should peak at 6.411 keV, while ferrihydrite should peak at 6.116 keV. Therefore, the peak at 6.405 keV could correspond to goethite, which is in agreement with previous studies on the biological removal of iron carried out by Madeleine et al. (2001). Since iron and manganese are intimately linked, we can see the manganese peak at a binding energy of 6.503 keV. Manganese oxides are generally expressed with the chemical formula of MnO_x, due to the multiple valence states exhibited by Mn. Therefore, it is reasonable to measure the average of the oxidation states of a manganese mineral (Eusterhues et al., 2008). In summary, Niton XL3T XRF Analyzer allowed us to detect the iron peak in the washing sludge. However, it would be necessary to use the Scanning Electron Microscope-Energy Dispersion Spectroscopy (SEM-SDE) to observe the exopolymers of bacteria, which have been immobilized in the filtration column.

This study contributes to the improvement of Senegalese groundwater treatment techniques in order to provide a technical solution to industrialists in the treatment of drinking water. Biological treatment has been studied for this purpose to remove the excessive quantities of iron in Senegalese groundwater.

As part of this work, we have carried out pilot tests in a continuous bioreactor. These tests made it possible to control the behavior of the bioreactor. In addition, the determination of the bed backwashing operating cycles showed that an operating cycle time of 1 day was sufficient for an initial concentration of 2 mg/L, while iron concentrations included 0.3 to 0.6 mg/L, no backwashing is necessary. Thus, the effect of varying the initial iron concentration improves the filter efficiency by approximately 5.84 to 4.47% under the specific experimental conditions and initial iron concentrations of 1.5 and 2 mg/L, respectively. During these tests on the pilot column, the biological oxidation of iron is a slow reaction, under the set experimental conditions, occurring in the reconstituted water to be treated.

The process has been tested on groundwater. The results show that the biological treatment gives results of reduction percentages of 39.3, 48.76 and 67.23%, respectively for the PK2, PS5 and KK5 boreholes. The difficulties of treatment for biological removal appear mainly at the level of groundwater with an increasingly important iron concentration.

The authors have not declared any conflict of interests.