Abstract

Among the methods of wastewater treatment, adsorption is the one that remains accessible because it is easy to implement. It uses less expensive, abundant and available adsorbent supports. Adsorption can be implemented through two types of processes: continuous adsorption and discontinuous adsorption. Continuous adsorption allows the treatment of large volumes of water compared to discontinuous adsorption, which is suitable for the treatment of small quantities of water [1]. This study is part of an approach to the removal of toxic dyes using the continuous mode adsorption method. Two dyes were used: crystal violet and methyl red. The adsorbent fixed bed of the column consists of biomass of previously deactivated lichens. The influence of some physico-chemical parameters on the column such as flow rate, lichen grain size, adsorbent bed mass and initial dye concentration were evaluated. The application of Bohart-Adams, Thomas and Yonn-Nelson models to the experimental data made it possible to predict the theoretical characteristics of the column. For a given dye, under the influence of a given physico-chemical parameter, the application of the Bohart-Adams model determined the saturation concentration of the adsorbent bed (N₀). The application of the Thomas model allowed to determine the theoretical adsorption capacity of the column (Q_the), and the Yonn-Nelson model allowed to predict the time (τ) necessary for the 50% breakthrough of the adsorbent bed.

1. Introduction

In recent years, Côte d'Ivoire has developed an industrial expansion, particularly in the plastic, rubber, wood, agro-industry, petrochemical, cosmetics, leather, paper and textile sectors ^{2, 3}. These industries use various dyes in their operations. Unfortunately, only 1% of these industries have effluent treatment plants ³. The effluents of most of these industrial units are thus discharged into the environment without any prior treatment. However, some of dyes contained in these waste waters are sometimes toxic or even carcinogenic to humans and also have harmful effects on the environment ⁴. Water pollution by dyes from these industrial effluents has therefore become a health and environmental problem. Numerous studies have focused on industrial wastewater treatment methods. Among these methods are electrochemical oxidation ⁵, ozonation ⁶, electrocoagulation ⁷, Fenton photo process ⁸, nanofiltration, reverse osmosis ⁹, etc. Some of these methods are expensive and inefficient for treating wastewater containing dyes. Several studies have revealed the limitations of these methods related to the production of toxic by-products ^{10, 11}. This has led the scientific community to move towards alternatives such as the adsorption method. Adsorption media can be natural or artificial. However, the use of natural adsorbents such as clay ¹², activated carbon ¹³, sawdust ¹⁴, peanut shells ¹⁵, orange peel ¹⁶ and banana peels ¹⁷ remains the cheapest and easiest to use. It is in this context that the present study was carried out with the aim of removing crystal violet (CV) and methyl red (MR), two toxic dyes, through a continuous adsorption column. The fixed bed of the column is a biomass of lichens. Lichens are plant species that result from the symbiotic association between an alga and a fungus. Lichens have been used as sorbents in several pollutant removal studies such as metallic trace elements ^{18, 19, 20, 21, 22, 23, 24}. However, there are few studies on dye removal by continuous adsorption using lichens as an adsorbent bed. This study will more specifically determine the theoretical characteristics of the proposed adsorption column. To do so, the Bohart-Adams, Thomas and Yonn-Nelson models were applied to the results of the evaluation of the influence of physico-chemical parameters such as flow rate, lichen grain size, lichen mass and initial dye concentration.

2. Material and Methods

The thalli of the lichen Parmotrema dilatatum were collected in the forest reserve of the LAMTO ecological station (5.02 ° C West and 6.13 °C North). This reserve is 174 km far from the city of Abidjan (Ivory Coast) and located on the Abidjan-Yamoussoukro axis, between Singrobo and Taabo, and stretches along the Bandama River. This space is a natural Park of 2500 ha, with a tropical climate, the average temperature per year is about 28.28 °C, the annual precipitation is around 1194 mm and the rate of humidity is higher than 58%.

Sampling is carried out on mature trees with diameters greater than 20 cm with bark that is not cracked, injured or slanted. Lichens collected are sent to the laboratory.

In the laboratory, lichens are detached from their substrates and then manually cleaned to remove soil, leaf, dust or insect debris. The thallus is then washed with distilled water. An appropriate amount of lichen is placed in an oven at 80 °C for 48 hours to deactivate them. The deactivated lichens are crushed and sieved.

Solutions of 500 mg/L of CV dye, and MR were prepared. We diluted the initial solutions in order to obtain diluted solutions.

The chemical structures and some characteristics of the dyes are given below (Figure 1).

The experimental device consists of a glass column 22 cm long and 8 cm inside diameter, a flow and storage vessel for the solution to be treated equipped with a valve with adjustable flow rate. A pump ensures the circulation of the solution to be treated in the column. The column contains a filter and reads adsorbent. Another container allows the treated solution to be recovered at the column outlet. (Figure 2).

The flow rate of the column was successively set at 0.07; 0.15 and 0.21 L/min. The effect of lichen grain size was examined using the grain sizes, [125 -250μm], [250 -500μm] and [500 -800μm]. As for the influence of the height of the adsorption bed, the study was carried out by successively setting the lichen mass at 20, 40, 60, 80 and 100g. The effect of the initial concentration of dyes was studied between 10 and 30 mg/L. The optical density of the treated solution was read using a UV/visible spectrophotometer. The amount of dye adsorbed per gram of lichen at equilibrium Q_exp (mg/g) was calculated using the following relationship ²⁵:

(1)

Where:

C₀ is the initial concentration of the solute (mg/L),

C_t the solute concentration (mg/L),

D flow rate (L/min),

t_total the saturation time,

m the lichen mass (g).

Q_exp is obtained by polynomial regression of Eq. (2), followed by its integration from t = 0 to t = t_s.

(2)

The characteristics of the adsorption column were determined by applying numerical models. Knowledge of the theoretical characteristics of the column allows a better description of the different stages of the adsorption process and to know the temporal concentration profile between the liquid and solid phases ²⁶. The most recommended models are the Adams-Bohart, Thomas and Yoon-Nelson models ²⁷. Each of these models can be used to determine different operating parameters of the fixed-bed column.

Bohart-Adams model ²⁸ assumes that there is only one adsorption layer in dynamic regime. The adsorption equilibrium in this case is achieved on each grain of adsorbent and the reverse desorption reaction is possible. The Bohart-Adams model predicts the bed saturation concentration (N₀).

The Bohart-Adams model is given by the following expression:

(3)

Where:

C₀ is the initial concentration of the solute (mg/L),

C_t the solute concentration (mg/L),

K_BA the kinetic constant of Bohart-Adams (L/mg.min),

N₀ the saturation concentration of the bed (mg/L),

v the flow rate of the solution (cm/min),

z is the height of the adsorbent bed (cm).

The linear form of the Bohart-Adams model is given by the following relation ²⁹:

(4)

Thomas model ³⁰ assumes that the adsorption process follows pseudo-second-order kinetics, with no axial dispersion during flow in the adsorption column ³¹. The Thomas model makes it possible to predict the theoretical adsorption capacity (Q_the) of the column. It is one of the most widely used models for the theoretical study of the performance of fixed bed adsorption columns.

Thomas model is given by the following expression:

(5)

Where:

C₀ is the initial concentration of the solute (mg/L),

C_t the solute concentration (mg/L),

k_Th the Thomas constant (L.mg^-1.h^-1),

m the mass of the adsorbent bed (g),

Q_the the theoretical adsorption capacity (mg/g),

D the flow rate (L.h^-1).

The linear form of this model is given by the expression:

(6)

Yoon-Nelson model ³² provides information on the time τ the required for 50% saturation of the column bed.

The Yoon-Nelson model takes the following form:

(7)

Where:

C₀ is the initial solute concentration (mg/L),

C_t the solute concentration (mg/L),

k_YN the Yoon-Nelson kinetic constant (min^-1),

τ the required for 50% saturation of the column bed.

The linearized form is given by the expression:

(8)

These three numerical models have in common the expression:

(9)

The function f(t) is specific to each model. The linear regression obtained from this expression makes it possible to determine the different parameters of each model. Among these parameters are the theoretical characteristics of the column.

Data analysis and linear regressions were performed alternately using Microsoft Excel 2010 software and STATISTICA 8.0.360.0.

3. Results and Discussion

The linear regressions of Eq. (9) obtained from the experimental results of the influence of dye flow through the fixed-bed column are shown in Figure 3.

The parameters of the Bohart-Adams, Thomas and Yoon-Nelson model relating to the study of the influence of the dye flow through the fixed bed column of deactivated lichens are recorded in Table 1.

For each dye, increasing the flow rate from 0.07 to 0.21 L/min, results in an increase in the Bohart-Adams speed constant (k_BA). The bed saturation concentration (N₀) decreases as the flow rate of the dye through the column increases. The highest values of the Bohart-Adams velocity constant (k_BA) are 15.1*10^-4 for CV and 35*10^-4 L/(mg.min) for MR. These velocity constants are obtained at the lowest flow rate (D = 0.07 L/min) with high correlation coefficients. The increase with flow of the constant k_BA, also called mass transfer coefficient, reflects the fact that the overall kinetics of the system is dominated by the mass transfer in the initial part of the bed of the column ³³. In their studies of phenol adsorption on a fixed-bed column of polymer resins, ³⁴ also found that increasing flow rate increased the k_BA constant but, contrary to our results, they observed an increase in the N₀ saturation concentration of the bed with increasing flow rate. However, ³⁵ in their work on antibiotic adsorption on a fixed bed column, showed that a low flow rate increased the bed saturation concentration (N₀) and decreased the Bohart-Adams k_BA rate constant.

The results in Table 1 show that the theoretical (Q_the) and experimental (Q_exp) adsorption capacities increase with the decrease in the flow rate of the dyes in the column. The Q_the values estimated by Thomas' model are close to Q_exp. The experimental data of the study are therefore well described by Thomas' model. Thus, with the flow rate D = 0.07 L/min, the quantities of CV and MR experimentally adsorbed are 14.29 and 8.87 mg/g respectively. Theoretically with the same flow rate, the quantities obtained are 14.52 mg/g for CV and 9.16 mg/g for MR. Whatever the dye, the values of the kinetic constant Thomas (k_Th) decrease with increasing flow rate. The same is true for the theoretical adsorption capacity. The high Q_the value at low flow rates is due to the slow contact, which promotes the diffusion ³⁶ of the dyes into the column bed. Conversely, the low Q_the value at high flow rate is due to insufficient diffusion of the dye molecules ³⁷. The good description of the experimental data by Thomas' model reveals that the adsorption process of the studied dyes follows second order kinetics ³⁸. The authors ³³ studying the removal of antibiotics on a fixed bed column also observed that the increase in flow rate induced a decrease in the theoretical adsorption capacity and an increase in the Thomas constant. The Thomas model was also used by ³⁴ in a fixed-bed column study. They found that the Thomas constant depended on the flow rate and that the theoretical amount of adsorption increased with decreasing flow rate.

The values of the Yoon-Nelson velocity constant (k_YN), the time (τ) required for 50% saturation of the column bed and the coefficient of determination are recorded in Table 1. For each dye, the k_YN values increase with increasing flow rate, while the time required for 50% saturation of the column bed decreases. Thus, for a flow rate of 0.07 to 0.21 L/min, k_YN increased from 55*10^-4 to 55.1*10^-3 min^-1 for CV and from 83*10^-4 to 35*10^-3 for MR. The time (τ) has increased from 1244.26 to 243.18 min for CV and 785.23 to 131.85 for MR. The decrease of the Yoon-Nelson velocity constant reflects the negative effect of the increased flow rate on the adsorption capacity ³⁹. Similarly, the time required to reach 50% bed saturation decreases with increasing column throughput due to the faster saturation of the bed ⁴⁰. A similar trend was observed by ⁴¹ in a study on the adsorption of amoxicillin in a fixed bed of activated carbon. Other authors ⁴² have also reported similar observations in simulated atenolol-carbon system failure data ⁴³.

Linear regressions of Eq. (9) obtained from the experimental results of the influence of lichen grain size in CV and RM adsorptions are presented in Figure 3.

Table 2 combines the Bohart-Adams, Thomas and Yoon-Nelson model parameters for studying the influence of lichen grain size on dye adsorption through the fixed-bed column.

The results recorded in Table 2 show that the fixed bed saturation concentration (N₀) decreases with increasing lichen grain size. However, the Bohart-Adams Velocity Constant (k_BA) increases with increasing lichen particle size. For grain size [125-250 µm], N₀ increased from 4331.95 to 2975.08 L/(mg.min) for CV and from 2733.78 to 2111.62 mg/L for MR. The high values of the Bohart-Adams constant (k_BA) in the same grain size inerval are 71*10^-4 L/(mg.min) for CV and 12*10^-4 L/(mg.min) with MR. The Bohart-Adams model is therefore perfectly described with small lichen grain sizes.

According to Table 2, the theoretical and experimental adsorption capacity values decrease as the grain size of the lichens that make up the fixed bed in the adsorption column increases. However, Thomas' kinetic constant increases with the diameter of the deactivated lichen grains. The values of the theoretical adsorption capacities (Q_the) predicted by Thomas' model are close to those obtained experimentally (Q_exp). Indeed, with lichen grain sizes between [125-250 µm]; [250-500 µm] and [500-800 µm]; the theoretical quantities of CV adsorption on the fixed bed are respectively 14.52, 13.42 and 9.98 mg/g while those obtained experimentally are 14.29, 13.35 and 9.83 mg/g. The values of the Thomas k_Th kinetic constant in CV adsorption range from 55*10^-5 to 71*10^-5 L/(mg.min) respectively with grain sizes from 125 µm to 800 µm. For MR, k_Th increases from 83*10^-5 to 12*10^-4 L/(mg.min). The good agreement between the experimental adsorption capacities and those theoretically predicted by Thomas' model shows that adsorption follows second-order kinetics ³⁸.

Analysis of the Yoon-Nelson model parameter values reported in Table 2, shows that lichen particle size influences the time τ, required for 50% saturation of the column bed. Particle size also influences the Yoon-Nelson kenetic constant (k_YN). An increase in grain size leads to a decrease in time τ and an increase in k_YN. When grain size varies from 125 µm to 800 µm, the time τ, drops from 883.58 min to 529.67 min with CV. For MR the time τ drops from 785.23 min to 606.53 min. The k_YN constant in the case of CV adsorption went from 55*10^-4 to 71*10^-4 L/(mg.min) and from 83*10^-4 to 12*10^-3 L/(mg.min) in the case of MR. The k_YN constant therefore increases with lichen grain size while time τ decreases. This can be explained by the fact that with small grain sizes there is a reduction in the diffusion path ⁴³ of the dye in the matrix. However, the authors ⁴⁴ in a study on the adsorption of ionic liquids through an activated carbon fixed bed column, observed high k_YN values with small activated carbon grain sizes.

The linear regressions of Eq. (9), obtained from the experimental results of the influence of lichen mass in the adsorption of dyes through the fixed-bed column, are shown in Figure 5.

The parameter values of the Bohart-Adams, Thomas and Yonn-nelson models determined from the study of the influence of the lichen mass, constituting the fixed bed of the column, are shown in Table 3.

The saturation concentration (N₀) of the bed and the kinetic constant (k_AB) are recorded in Table 3. The saturation concentration increases with the mass of the adsorbent bed. The kinetic constant (k_BA) decreases with increasing lichen mass in the column. Thus, for CV, the saturation concentration increased from 2984.39 to 6362.67 mg/L for a mass change of 40 to 100g of deactivated lichens. For MR, the N₀ increased from 1893.15 mg/L to 4351.05 mg/L. The increase in the mass of the adsorbent bed resulted in a decrease in the kinetic constant k_BA from 10.4*10^-4 to 17*10^-5 L/ (mg.min) for CV and from 15*10^-4 to 27*10^-5 L/ (mg.min) for MR. The authors ³⁴ applied the Bohart-Adams adsorption model to describe the adsorption of phenol on polymer resins in a fixed-bed column. They also found that the saturation concentration (N₀) decreased as the mass of the adsorbent bed increased. However, they observed an increase in the k_BA constant with increasing adsorbent bed mass.

Analysis of the values in Table 3, obtained by application of Thomas' model, shows that the theoretical (Q_the) and experimental (Q_exp) adsorption capacities of the CV and MR dyes are close and increase with the mass of the column bed. In the case of Q_exp, that of CV rose from 7.13 to 21.6 mg/g and that of MR from 4.75 to 17.33 mg/g. In the case of Q_the, it rose from 7.501 to 25.586 mg/g in the case of CV and from 4.758 to 17.497 mg/g in the case of MR. Table 3 further reveals that, as the adsorbent bed mass increases, Thomas' rate constant decreases. Such a trend was observed by ³⁴ in the case of phenolic adsorption of vinasse wines on a fixed bed column of polymer resins. Other authors ⁴⁵ also applied Thomas' model in the case of tetracyline adsorption in a modified silica fixed-bed column. They also found that the adsorption capacity increased and k_Th also decreased with the height of the column bed.

Examination of Table 3 shows that the increase in the mass of the adsorbent bed induces an increase in the prediction time (τ) required for 50% saturation of the column bed and a decrease in the kinetic constant k_YN. Thus, considering a lichen mass varying from 40 to 100 g, the time τ increases from 428.61 to 3655.1 min in the case of CV treatment and from 271.89 to 2499.5 min in the case of MR. The variation in mass also leads to a decrease in the rate constant (k_YN) from 10.4*10^-3 to 17*10^-4 L/(mg.min) for CV treatment and from 15.4*10^-3 to 27*10^-4 L/(mg.min) for MR. The researchers ⁴¹ have recently applied the Yoon-Nelson model for the adsorption of amoxicillin on a fixed bed activated carbon column. The results of their work showed that the prediction time (τ) for 50% absorption of amoxicillin increased and that the Yoon-Nelson kinetic constant decreased with increasing bed height. The authors ³⁴ in a study on the adsorption of phenol through a fixed bed column of polymer resins also observed an increase in time (τ) and a decrease in k_YN with increasing mass of the adsorbent, constituting the fixed bed of the column.

Figure 6 shows linear regressions of Eq. (9) obtained from the experimental results of the influence of the initial dye concentration on lichen adsorption capacity.

Table 4 groups the parameters of the Bohart-Adams, Thomas and Yonn-Nelson models from the study of the influence of initial dye concentration.

Values of Bohart-Adams model parameters in the removal of CV and MR at different initial concentrations are presented in Table 4. Analysis of these values shows that increasing dye concentration affects the bed saturation concentration (N₀) and the Bohart-Adams kinetic constant (k_BA). N₀ increases with increasing dye concentrations, while k_BA decreases. Thus, a change in CV concentration from 10 to 30 mg/L resulted in an increase in N₀ from 4331.95 to 5641.25 mg/L. As for MR, a change in concentration from 10 to 30 mg/L increased N₀ from 2733.78 to 4595.47 mg/L. The authors ³⁵ applied the Bohart-Adams model to the adsorption of two drugs (tetracyline and chloramphenicol) through a fixed-bed bamboo charcoal column. These authors reported that as the initial drug concentrations at the column inlet increased, N₀ also increased while k_BA decreased. Similarly, the authors ³⁴, observed that the kinetic constant (k_BA) decreased with increasing phenol concentration and that the saturation concentration (N₀) also increased with the initial concentration.

The values of the theoretical capacity (Q_the), the experimental capacity (Q_exp) and the Thomas' kinetic constant (k_Th) are contained in Table 4. From this table, it can be seen that, regardless of the dye, Q_the is close to Q_exp. The latter two parameters increase as the initial concentration of the dye increases; thus, for a change in initial dye concentration from 10 to 30 mg/L, Q_exp increases from 14.29 to 18.3 mg/g and Q_the increases from 14.52 to 18.9 mg/g for CV. For MR, Q_exp increases from 8.87 to 15.25 mg/g and Q_the increases from 9.16 to 15.4 mg/g. However, k_Th decreases with increasing initial dye concentration. A decrease in k_Th and an increase in Q_the and Q_exp with increasing initial dye concentrations can be explained by the fact that at lower concentrations the mass transfer is slower and thus improves the adsorption capacity ⁴⁰. A similar evolution of Q_the, Q_exp and k_Th was observed by ⁴⁶ in a study of the adsorption of salicylic acid through a fixed bed column. These authors observed that k_Th values decreased while theoretical and experimental adsorption capacities increased with increasing initial salicylic acid concentrations.

From the results in Table 4, it can be seen that increasing the initial dye concentration results in a reduction in the time (τ) required to saturate 50% of the column bed and an increase in the Yoon-Nelson kinetic constant (k_YN). The time (τ) is reduced from 1244.27 to 540.17 min in the treatment of CV and from 785.23 to 440.03 min in the case of MR, for an initial dye concentration ranging from 10 to 30 mg/L. As for (k_YN), it decreased from 55*10^-4 to 10.1*10^-3 L/(min.mg) for CV and from 83*10^-4 to 11.9*10^-3 L/(min.mg) for MR.

The time (τ) decreases with increasing initial dye concentration due to faster saturation of the column bed ⁴⁰. The authors ⁴¹ applied the Yoon-Nelson model in the adsorption of amoxicillin through a fixed bed activated carbon column. The results of their study revealed that the time (τ) decreased while k_YN increased with increasing initial concentration of amoxicillin. Such trends were also observed by ⁴⁷, in the fixed bed adsorption of doxycycline on an akagan-carbon composite; and by ⁴⁸ in the fixed bed adsorption of ibuprofen on raspberry leaf charcoal.

4. Conclusion

This study made it possible to determine the theoretical characteristics of a column in the elimination of violet crystal and methyl red, by adsorption in continuous mode on a fixed bed of deactivated lichens. The parameters of the Bohart-Adams, Thomas and Yonn-Nelson models vary according to the dye flow rate in the column, the size of the lichen grains, the mass of the adsorbent bed and the initial concentration of dyes.

The theoretical characteristics obtained, that is to say, the saturation concentration (N₀) of the adsorbent bed, the theoretical adsorption capacity (Q_the) and the time (τ) required for 50% saturation of the column bed, are better at low dye flow rates. These theoretical characteristics are also better with the decrease in the size of lichen grains, the increase in the mass of the adsorbent bed and with high initial concentrations of dyes.

However, it would be interesting to determine these theoretical characteristics in a real industrial effluent treatment situation.

References