Several physical/chemical treatments described by [4] [13] were carried out on the raw brick, briefly: 1) it was broken into grains and sieved to sizes ranging from 0.7 to 1.0 mm; 2) these particles were afterwards leached with a 6 M-HCl solution at 90˚C for 3 h; and 3) finally a deposition of FeOOH on to HCl-treated brick was performed by precipitation of iron(III) by using a 0.25M Fe³⁺ ions solution with NaOH solutions (6 M and 1 M) at pH 7.

All chemicals used for the study were analytical grades of ferric nitrate [Fe(NO₃)₃, H₂O], ferrous chloride [FeCl₂, 4H₂O], sodium hydroxide and hydrochloride acid supplied by DISLAB (France). The stock solutions of iron(III) (0.25 mol/L) and iron(II) (10 mg/L) were prepared by dissolving the corresponding salts in Milli-Q water before performing the experiments. It is worth noting that Fe²⁺ ions initially present in the prepared solution (10 mg/L) remained at oxidation state II under our physicochemical conditions (pH ≈ 4.5; Eh = 347 mV vs Ag/AgCl ( (KCl) = 3 mol/L)); [O₂] = 5.77 mg/L). Reference electrode has been calibrated whit a redox standard solution (Methrom U (mv) = +250 ± 5 (20˚ pH = 6.994)). The thermodynamic stability of ferrous ions in our solution was further confirmed theoretically by examining the pɛ-pH diagrams for the iron-water system [18] .

Microphotographies of representative brick specimens after precipitation of iron oxyhydroxide were recorded by using an environmental scanning electron microscope (ESEM, Quanta200 FEI). Different surface areas ranging from 0.5 to 3.5 mm² were targeted on brick grain sand examined by ESEM/EDS (ESEM, model: QUANTA-200-FEI, equipped with an Energy Dispersive X-Ray Spectrometer EDS X flash 3001 and monitored by QUANTA-400 software elaborated by Bruker). EDS measurements were carried out at 20 kV at low vacuum (1.00 Torr) and the maximum pulse throughput was 20 kcps. Micro-observations along different cross-sections of FeOOH-coated brick samples were also performed by ESEM/EDS.

The powder X-ray diffraction (XRD) was performed by using a Bruker diffractometer (model: Endeavor D4) equipped with a PSD/Lynx Eye detector. The powdered sample was scanned from 10˚ to 8˚ 2θ with increments of 0.0157˚ 2θ and a number of 4456 steps using a Ni-CuKα radiation (the recording time was ≈ 7.2 h). The Cu tube operated at 35 kV and 30 mA. To detect good XRD diffractogram of 2-line ferrihydrite associated with brick, the fraction of brick particles with sizes ≤ 63 µm was recovered and chemically treated according to the procedure described by [14] [15] : A thinner powder was then used in this work to precipitate higher amounts of iron oxyhydroxides on to material surfaces.

ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectroscopy; model Varian Pro Axial View) was used to analyze the soluble iron effluent samples that were collected from the exit of the bottom column at different time intervals.

Continuous flow adsorption according to the experiments described by [13] : was conducted in a fixed-bed glass column with an inner diameter of 14 mm, a height of 30 cm, and a medium porosity sintered-pyrex disk at its bottom in order to prevent any loss of material. Four-bed depths of 10 cm (16 g), 15 cm (23 g), 20 cm (30 g), and 25 cm (38 g) were investigated at a constant flow rate of 10 mL/min, and a constant bed depth of 15 cm was examined at variable rates (3, 5, 8, and 10 mL/min). Before being used in the experiments, 500 ml of Milli-Q water was passed through the column for three reasons: 1) to remove any unbound and thin particles/iron oxides/hydroxides; 2) to check the absence of soluble iron at the outlet; and 3) to confirm the stability of the FeOOH coating on brick pellets. The initial iron(II) concentration of the influent was 10 mg/L. The schematic diagram of the fixed bed column reactor used is shown in (Figure 1).

Fe²⁺ ions solution was pumped through the column at a desired flow rate by means of a peristaltic pump (Labo Moderne France Type KD1170) in a downflow mode. Depending upon our various study column experiments, measured pH was found to be around 4.5 in the influent and to vary in the effluent from 8.6 at the beginning to 5.5 at the end. Effluent samples were collected from the

exit of the bottom column at different time intervals and analyzed for soluble iron using ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectroscopy; model Varian Pro Axial View). The flow to the column was continued until the effluent Fe(II) concentration at time t (C_t) reached the influent Fe(II) concentration (C₀): C_t/C₀ ≈ 0.98. The performance of packed bed was described in the present work through the concept of the breakthrough curve. The shapes of the breakthrough curves observed during column experiments were studied at variable flow rates and bed depths, and mathematical models were applied to our system.

The breakthrough time (t_b) corresponding to 5% of initial concentration, the time t50% corresponding to 50% of initial concentration and the exhaust time (t_e) corresponding to the time required to achieve 90% of the inlet concentration were determined. The analysis of the breakthrough curve was done using two models, the prediction of theoretical breakthrough profiles [19] and the bed depth/service time (BDST) model [17] . The BDST model was extended to the prediction of breakthrough and exhaust times of the columns operated at different flow rates and also varied bed height once the slope at a given flow rate and bed height was determined.

After deposition of ferrihydrite by precipitation of iron oxyhydroxides on the surface of brick grains according to the procedure described by [13] , the micro-observations along different cross-sections of FeOOH-coated brick samples show a deposition of the heterogeneous oxyhydroxides in plate form which is characteristic of gel drying (Figure 2(a)).

The profile analysis of the surface of a brick grain after precipitation over a distance of 100 μm (Figure 2(b)) shows that iron and silicon are anti-correlated while iron and aluminum are correlated (see Figure 2(c)).

This shows that the FeOOH deposited on the grains of the brick adsorbs preferentially on metakaolinite and little on quartz.

The raw brick obtained from Bangui region in Central African Republic used in this study was previously characterized by X-ray diffraction and chemical analysis as described before [13] [14] Average mineralogical composition was found to be: 61 - 65 wt% quartz; 21 - 25 wt% metakaolinite; 3 - 4 wt%; illite; ≤ 4 wt%; iron oxides/hydroxides; and ≤2 wt% feldspar + mica + biotite. The XRD analyses of FeOOH-coated brick have been extensively reported in previous studies [13] It shows in the diffractogram of iron oxyhydroxides coated brick two broad maxima at 34˚and 61˚ 2θ corresponding to d-spacings of 1.52˚A and 2.63˚A, typical of two-line ferrihydrite (Figure 3).

During the crossing of the column, iron(II) ions are adsorbed on the fixed bed and the solution is exhausted. The iron(II) metal ions in solution are adsorbed at the first adsorbent layer called the exchange zone. During the time t < t_b, the outlet concentration of the iron(II) is almost zero. The exchange zone becomes saturated and the exchange zone progressively moves slowly towards the bottom of the adsorbent bed (Figure 1). Finally, the exchange zone will reach the bottom of the column at t = t_b and the solute concentration in the outlet effluent begins to increase at t > t_b and reaches its initial value at t = t_e.

In order to bring out the fixed bed column, several parameters influencing the breakthrough time were performed to determine the optimal operating conditions of treatment using ferrihydrite-coated brick fixed bed column.

To obtain the fixed bed maximum capacities, the effect of flow rate was investigated at a constant bed height of 15 cm. It can be seen in (Figure 4) the plots of comparative normalized iron(II) concentration, C_t/C₀, against the time passed through a 14 mm diameter column of the modified brick at different flow rates. Also, from the plots in (Figure 4) sorption data were assessed, and particularly, amount of total iron-sorbed, maximum iron uptake per gram of brick and total removal percent with respect to volumetric flow rate are listed in Table 1.

As a whole, it was noticed mainly three features: 1) metal-uptake values remained relatively constant with flow rate (0.338 ± 0.010 mg/g for F = 3 - 10 mL/min); 2) sharper breakthrough curves were obtained at higher flow rates (the ∆c/∆t ratio which is defined in Table 1, varied from 0.069 mg/L min for F = 3

mL/min to 0.174 mg/L min for F = 10 mL/min); and 3) the breakpoint time (t_b) decreased with increasing flow rates and at a constant mass of brick inside the column: m = 23.4 ± 0.4 g (from t_b = 151 min for F = 3 mL/min to t_b = 50 min for F = 10 mL/min). It should further be noted that a better Fe(II) adsorption performance of the study brick was obtained at the lowest flow rates, due to higher residence time.

The breakthrough curves of iron(II) adsorption which were obtained in a 14 mm diameter column at different bed heights with a constant flow rate of 10 mL/min are given in (Figure 5). For this column (14 mm diameter), the bed height was about 10, 15, 20 and 25 cm, which corresponded to brick masses of 16, 23, 30, 38 g, respectively.

It was found that a higher bed height or an increase in adsorbent weight inside the column led to higher quantities of adsorbed Fe(II) in the column reported in Table 1, due to the involvement of more adsorption brick sites in interaction with the effluent. Obviously, the maximum capacity of iron-uptake per gram of brick remained relatively constant with increasing bed height (as reported in Table 1). This result could be attributed to the number of up-take sites per gram of brick which remained constant when increasing bed height.

The time required for the exchange zone to move the length of its own height up/down the column (t_z) once it has become established is [18] [19] [20] :

t z = V s F (5)

where Vs is the total volume (mL) of water treated between t_b and t_e and F is the

flow rate (mL∙min⁻¹).

The time required for the exchange zone to become established and move completely out of the bed (t_e) is:

t e = V e F (6)

where Ve is the total volume of iron(II) spiked water treated to the point of exhaustion (L).

The rate at which the adsorption zone is moving up and down through the bed is:

U Z = h Z t Z = h t e − t f (7)

where h_z is the height of exchange zone (cm); h is the total bed depth (cm); t_f is the time (min) required for the adsorption zone to form initially.

The expression for the height of the exchange zone (h_z) is:

h Z = h ( t Z ) t e − t f (8)

The value of t_f can be calculated as:

t f = ( 1 − f ) t Z (9)

where f is the fraction of iron(II) adsorbed when the adsorption zone is completely saturated after equilibrium, is given by:

f = S Z S max = ∫ V b V e ( C 0 − C t ) d V C 0 ( V e − V b ) (10)

where C₀ is the initial concentration of iron(II) in water (mg/L); C_t is the concentration of solute iron(II) at any instant in the effluent (mg/L); V_b is the total volume of iron(II) spiked water treated to the point of breakthrough (L); S_z is the amount of iron(II) that has been removed by the adsorption zone from breakthrough to exhaustion (mg) and S_max is the amount of iron(II) removed by adsorption zone if completely exhausted (mg).

The percentage of total column bed saturated at breakthrough is:

% s a t u r a t i o n = h + ( f − 1 ) h Z h × 100 (11)

It was noted that the moving rate of the exchange zone (U_Z) increases with the flow rate. U_Z ranges from 0.072 to 0.214 cm/min when the flow rate increases from 3 to 10 ml/min (Table 2). These values are in agreement with the experimentally determined ratio Δ(C_t)/Δt which decreases with the flow rate. On the other hand, it shows in Table 2 that the height of the exchange zone remains almost constant at all the rates of the filtration. Therefore it was found that the exchange zone does not depend on the height of the fixed bed and the filtration rate but probably on the affinity of the solute with the adsorbent. The variation

of the percentage of saturation as a function of the flow rate is not significant to affirm an influence of the variation of flow rate on the height of the exchange zone.

The mathematical evaluation of the basic adsorption column design parameter using various bed heights permitted us to show that the moving of the adsorption zone in the adsorbent bed is independent of the bed height and remained constant (U_Z = 0.208 ± 0.006) cm/min when increasing bed height. The values of U_Z are close to those experimentally obtained for the ratio Δ(C_t)/Δt between the breakthrough time (t_b) and the exhaustive time (t_e). This ratio also remains constant and equals 0.17 ± 1 mg/l/min and corresponds to the variation of the concentration of iron(II) adsorbed on the support per unit of time.

It was found when the bed heights of column increased to 10 from 25 cm, the height of the exchange zone in the column also remains constant, h_Z = 10.3 ± 0.6 cm.

The time required for the adsorption zone to move vertically up the column (t_Z) and the time required for the adsorption zone to become totally saturated were also constant (see Table 2), in agreement with the moving rate and the height of the adsorption zone which were invariable when the heights of column bed varied from 10 to 25 cm Between the breakthrough time (t_b) and the exhaustive time (t_e) when the heights of column bed varied to 10 from 25 cm, the percentage of the total column saturation at breakthrough varied from 41.59 to 73.31. This increase of the saturation percentage is in agreement with the theoretical data from Equation (3).

A number of mathematical models have been developed for use in design. Some investigators attempted to simulate column performances by testing mathematical models from data collected during laboratory experiments in order to design pilot-scale columns [21] [22] [23] with mathematical models. Among them, differential Bohart-Adams equations [18] [24] [25] were successfully applied to liquid-solid systems by accounting for the dynamics of the study adsorption process. Briefly, the decrease with time of the adsorption capacity, “No”, of the adsorbent was found to be proportional to the adsorption rate, K_ads, as follows:

∂ N 0 ∂ t = k a d s N 0 C (12)

Hutchins [26] proposed a simplified equation based on the above surface-reaction-rate theory proposed by Bohart and Adams (1920): it was presented as a linear relationship between the column bed depth (Z (cm)) and the service time (t (min)):

t = N 0 C 0 υ Z − 1 k a d s C 0 ln ( C 0 C t − 1 ) (13)

where N₀ represented the dynamic bed capacity (mg/L); ν was the linear flow rate (cm/min) which was defined as the ratio of the volumetric flow rate F (cm³/min) to the cross-sectional area of the bed, S (cm²); K_ads. was the adsorption rate constant (L/mg min); and C₀ and C_t corresponded to the influent Fe²⁺ concentration and the effluent Fe²⁺ concentration at time t, respectively.

At a time (t_50%) at which Fe(II) concentration in the effluent reached 50% of the influent Fe(II) concentration, BDST Equation (14) can be simplified as:

t 50 % = N 0 Z C 0 υ (14)

The critical column height (Z₀) is the minimum column height necessary to produce C_t < C_5% of the initial concentration of the effluent.

The time t = 0 was determined in Equation (13).

Z 0 = υ k a d s N 0 ln ( C 0 C 0.05 − 1 ) (15)

During the experiments, three particular values of the service time were studied: t_b, t_50% and t_e, which correspond to 5%, 50%, and 90% of initial concentration in the effluent at the outlet of the column. The variation of these three service times V_s the height of the bed column permitted us to get three lines in agreement with the Hutchins equation (Figure 6). Which showed the depth vs service time for t_b = 0.5%, t_50% = 50% and t_e = 90% saturation of the columns. The equations and correlation coefficients of these lines are as follows:

t b = 5.18 Z − 22.9 , R 2 = 0.98 (16)

t 50 % = 5.38 Z − 7 , R 2 = 0.98 (17)

t e = 5.62 Z + 20.9 , R 2 = 0.98 (18)

From the slopes these lines t_b and t_e are parallel and the horizontal distance

between the lines was found to be 10.71 cm. This horizontal distance is the height of the exchange zone in agreement with theoretically value that was averaged to be 10.3 ± 0.6 cm (Table 2).

The slopes of these 3 lines are similar and allow from Equation (13) to calculate from the 3 times (t_b, t_50% and t_e) an average value of dynamic capacity N₀: 345 ± 15 mg/L. The minimum column height (Z₀) necessary to produce an effluent concentration of C_t < C5% was showed in Table 3. It can be seen Z₀ remained nearly constant at a given flow rate of 10 mL/min and at bed height varying from 10 to 25 cm. This meant that Fe(II) adsorption onto brick pellets ought to be efficient only if studied bed heights Z was >Z₀.

From this value N₀, the adsorption constant K_ads can be reached. This constant calculated from the different times t_b increases slightly with the height of the bed (Table 3). The reason for weak increases observed for K_ads might be due to physical column parameters like wall effects.

By plotting the variations of t_b, t_50%, and t_e Vs 1/ʋ with the ʋ was a linear rate in cm/min of liquid flow calculated from the flow rate (Figure 7).

t e = 304.75 x + 3.351 , R 2 = 0.92 (19)

t 50 % = 405.8 x + 1.2612 , R 2 = 0.99 (20)

t b = 625.39 x + 9.8837 , R 2 = 0.98 (21)

Regression coefficient values: show that the results obtained are well representative of Equation (14). From the slope and the intercept these parameters N₀, K_ads and Z₀ were calculated and shown in Table 3.

By taking into account the values of the initial concentration and linear rates of filtration, the dynamic bed capacity was found to be N₀ = 335 mg/l. This value is very close to N₀ obtained by varying the height of the bed, i.e. N₀ = 345 mg/l. The adsorption rate constant K_ads does not change significantly with the rate and is 32.4 ± 2 mL/(mg/min). The critical bed depth, Z₀, increased from 0.83 to 3.25

cm with increased flow rate from 3 to 10 mL/min at a given bed height of 15 cm. The Fe(II) adsorption onto brick was efficient because bed heights Z > Z₀.

To compare, the experimental data shows that the values of height of the exchange zone was in agreement with theoretical data using BDST model. Both curves revealed a good correlation for the treated iron system.