· Presentation of the study area and collection of site data

The present study was carried out on two sites: the highlands of West Cameroon, precisely in the town of Foumbot, Nkouonke village in Noun, where andosols were sampled (the NK1 under fallow soil, the NK2 bare soil and the NK3 cultivated soil); then in South Cameroon where Ferralsols were sampled (the cultivated soil MIN2, the soil under fallow MIN3 and the bare soil MIN1). The choice of these soils is justified by the high level of agricultural activity in these areas.

The pesticide paraquat, which is a herbicide, was used. The choice of this pesticide is justified by its massive use in peasant and industrial crops in Foumbot and Ebolowa.

The soil samples were characterised by the analytical method in accordance with the international standards in force at the Research Unit for Soil Analysis and Environmental Chemistry of the University of Dschang, as recommended by Pauwels et al. The different analyses of the soil samples were carried out according to the classical procedures of the soil, plant and water laboratory of the Faculty of Agronomy and Agricultural Sciences (FASA). The present research work was carried out in the field for the collection of soil samples and in the laboratory for the different analyses and adsorption studies.

Seven study sites representing typical characteristics of the study area and based on landscape, cropping system, organic matter composition, clay and iron content were selected. The physical and chemical properties of each site are given. After the preliminary adsorption studies, soils with low retention capacity were eliminated and only four soil samples were selected for the sorption experiments. All these soil samples were collected from the surface horizon (0 - 25 cm), crumbled, spread in a thin layer, air-dried in a ventilated area, free from contamination on plastic trays for about seven days. The samples were stirred daily to promote drying and then moderately crushed in a porcelain mortar with a porcelain pestle, without breaking up the sands and stones. The whole was sieved with a 2 mm mesh sieve and the sieve (fine soil) was kept in sealed and labelled plastic bags.

The soil profiles of the different soil samples from Foumbot (NK1, NK2 and NK3) and Ebolowa (MIN1, MIN2 and MIN3) were produced according to the description of soil profiles proposed by the FAO (FAO, 1990). For the classification of soils, the determination of the colours of the horizons of the different profiles is done by a “Munsell” colour chart. The soil profile of Foumbot FBT located in the Foumbot Caisserie district is not included in the statistical analyses. This site was studied for simple observation.

The sampling and preparation of the soil samples was determined according to the methods recommended by Pauwels et al. The selected locations were well cleaned. Soil profiles of about 150 cm × 150 cm surface area and 150 cm depth were dug at the Nkouonke and Foumbot Caisserie sites in West Cameroon and at Minkolmingon (MIN1), Tyele (MIN2) and Mekoto (MIN3), in Central and South Cameroon. Two types of sampling were carried out:

- Sampling of the different horizons along the soil profile at different points, which consisted of collecting only those horizons that could be used to classify the soil. Soil horizons were identified by colour differences using a “Munsell” colour chart, texture, root penetration and presence of gravel. Disturbed 2 kg samples from the central part of each horizon were taken with a clean knife, starting from the lower horizons (bottom up). They were then placed in a well-labelled plastic package and transported to the laboratory.

- Collection of different depth samples along the soil profile. The aim was to study the vertical storage and mobility of substances in the soil. Disturbed samples of 2 kg depth (0 - 25 cm, 25 - 50 cm) from each profile were taken with a clean knife, starting from the lower horizons (bottom up), then put in a well labelled plastic package and transported to the laboratory.

The depth of the soil to which the plant draws water and nutrients (topsoil) depends on the depth of rooting, which in turn is determined by the plant species and edaphic constraints. Routine analysis is carried out on surface samples (0 - 25 cm). These were collected and prepared for physicochemical analysis and adsorption studies. A 2 kg mass of surface sample was hand augered three times from the Nkouonke and Foumbot Caisserie sites in West Cameroon and the Minkolmingon, Tyele and Mekoto sites in Central and South Cameroon for physicochemical analyses and adsorption studies. Twenty-five soil samples were taken according to an experimental protocol proposed by FAO. After the preliminary studies, which consisted of the first adsorption experiments, twenty-one soil samples that showed a very low adsorption rate were eliminated; only four that showed an average adsorption rate were retained for further research. The soil samples selected were mainly the surface layers with a depth of 0 to 25 cm.

The analysis of this diffractogram shows that the dominant peaks are those of α-quartz SiO₂ silica (2θ = 20.8˚; 26.7˚; 50.3˚ and 60.1˚). According to the literature, these distances are in agreement with those found by [2] [17] . The well-resolved peak at 12.4˚ is attributed to aluminosilicate plates containing no metallic elements, which corresponds to the kaolinite Si₂Al₂O₅(OH)₄ (d = 7.13 Å). According to the work of [18] , some peaks corresponding to kaolinite can be observed at 2θ = 24.9˚ and 38.7˚ for the (002) reflection planes. Other peaks, less intense, indicate the presence of goethite (2θ = 21.2). The broad diffraction peak at 33.3˚ may be from 2 phases, goethite α-FeOOH (reflection (130)) or hematite α-Fe₂O₃ (reflection (104)) [19] . Goethite, which is the free iron, is present outside the clay systems.

IR analysis of soil samples

Figure 3(a) and Figure 3(b) show the IR spectra of the NK1 fallow and NK3 cultivated soils. Overall, it appears that the IR spectra of these different soils show the same pattern: these spectra have a very similar band distribution in the mid-infrared region for the different types of materials. This is proof that all materials have mostly identical functional groups. According to [20] , these observations could explain why this technique is not sensitive enough to clearly illustrate the different types of defects in all these materials. Using the results of the literature on IR spectroscopic characterization of soils, vibrational bands have been assigned:

Some characteristic peaks can be highlighted, the presence of a broad band in the region of 3690 - 3619 cm⁻¹ is assigned to the stretching vibration of the internal O-H hydroxyl groups of kaolinite [21] . There are absorption bands observed around 1630 and 1400 cm⁻¹ in Figure 3(a) and Figure 3(b) which are attributed to the molecular O-H deformation vibration of water and the C-H elongation vibration of organic matter respectively [22] . Absorption bands around 1000 and 910 cm⁻¹ observed in the above mentioned figures are assigned to the Si-O elongation vibration and the Al–AlOH elongation vibration bands of kaolinite respectively. The doublet observed in the spectra of the same samples in the 745 - 790 cm⁻¹ region can be attributed to the intertetrahedral Si-O-Si bond of quartz [21] .

The study of the contact time makes it possible to determine the time required to reach the adsorption equilibrium between the adsorbent and the adsorbate. To determine this, the material and the pollutant are brought into contact for various periods of time. This time indicates the end of the adsorption process. The adsorption equilibrium reached between the soil and the pesticide will make it possible to determine the different points constituting the isotherm used. The results collected during this study are given in Figure 4.

Analysis of these curves shows that they generally show the same pattern. The rapid adsorption of paraquat was observed during the first fifteen and twenty minutes respectively for the cultivated soil NK3 and the fallow soil NK1 until about forty minutes, which is the equilibrium time for these different soils to reach the plateau. The slope shows that there is a high affinity between the different soils and paraquat at the beginning and a saturation of adsorption sites afterwards. The NK3 cultivated soil adsorbs more and stabilizes quickly. We can say that there are two types of adsorption, one with a very steep slope and stabilizes; then the others that adsorb less and continue to adsorb. The first part characterized by a high adsorption rate is due to the fact that the number of adsorption sites initially available on the soil is high, allowing paraquat to reach them easily. The second part shows that as the contact time increases, the number of free sites on the soils decreases and the non-adsorbed molecules are assembled on the surface of the material; this limits the adsorption capacity (Figure 4).

The determination of the time required to reach adsorption equilibrium requires the kinetic study of adsorption. This study also makes it possible to determine, in a comparative manner, the quantities of pesticides adsorbed as a function of time for a given initial concentration. Finally, the kinetic study of adsorption provides information on adsorption mechanisms by providing them with information. In order to study the different kinetic parameters of the adsorption of pesticides, the kinetic models of pseudo-first order, pseudo-second order and Elovich were applied to the experimental data obtained during the study of the influence of contact time on adsorption. The linear regressions relating to each of these models were plotted and the constants obtained are presented in Table 3.

1) Pseudo-first-order, pseudo-second-order and Elovich kinetic models

The adsorption kinetics of paraquat by the three soils are presented in Figures 5(a)-(c) and the summary of the different constants obtained on the kinetic models of paraquat adsorption on the soils is presented in Table 3. The various adsorption kinetics were obtained for an introduced paraquat concentration of 10⁻⁴ mol·L⁻¹.

k₁ (min⁻¹), q_e (µmol/g), q_exp (µmol/g), k₂ (g µmol/ min), h (µmol/g/min), t_1/2 (min), α (µmol/g/min), β (g/µmol).

Table 3 shows that the adsorption of paraquat by the soils of Foumbot NK1, NK3 and Ebolowa MIN1 is better described by the pseudo-second order kinetic model since the q_e values obtained from this model are close to the experimental values. Observation of this table shows that the correlation coefficients for this model are close to unity. This proves that this model is the most suitable for interpreting these experimental data. This observation is regular for heavy metals or for organic compounds [23] [24] . These good correlations also indicate that chemisorption is the limiting step of the process [24] [25] .

The half-reaction time is the time after which half of the substance initially introduced is degraded. This time gives an indication of the persistence of the parent molecule in water, air or soil; but, it does not provide information on the metabolites. The half-reaction times indicate that the kinetics of pollutant accumulation is faster on the surface of fallow soil NK1 (t1/2 = 11.30 min.), followed by cultivated soil NK3 (t1/2 = 19.94 min.) and finally the bare ground of Ebolowa MIN1 (t1/2 = 264.05 min.). This observation shows that the soil under NK1 fallow seems more active or more available for paraquat fixation. There would be a correlation between the densities of the materials and these half-reaction times because the observation made is that they increase with the density of the materials used; which means that the diffusion of paraquat towards the fixation sites is favored in porous materials [24] .

The cultivated soil NK3 adsorbs at equilibrium a quantity of paraquat of 34.620 µmol/g followed by the soil under fallow NK1 with 31.870 µmol/g and finally the bare forest soil of Ebolowa MIN1 with 28.720 µmol/g. The soils, cultivated NK3, under fallow NK1 and bare MIN1 have respective organic matter rates of 5.07%, 8.61% and 3.78%. The quantities of paraquat retained with these different soils are not all the greater when the organic matter content of the soil is high. The differences in organic matter content are not clearly reflected in the quantities retained. Organic matter does not seem to be the only factor involved in the retention of paraquat with these different soils. This classification of soils according to the importance of the quantities of paraquat retained does not make it possible to find the same classification with the rate of organic matter in the soil.

The purpose of adsorption isotherms is to study the evolution of adsorption at equilibrium as a function of pesticide concentration. The isotherm curves represent the amount of pesticide adsorbed (q_e) as a function of the residual concentration of pesticide at equilibrium (C_e). In order to determine the maximum amount of pesticide that could be adsorbed onto the soil, the adsorption isotherm was performed by varying the initial pesticide concentration while keeping the other parameters constant. The adsorption isotherms of paraquat concentration 2 × 10⁻⁴ mol·L⁻¹ on NK1, NK3 and Min1 soils are obtained for introduced paraquat concentrations between 10⁻⁴ and 10⁻³ mol·L⁻¹ corresponding to 3.5 to 7.0 µmol·L⁻¹. The contact time between soil and paraquat corresponds to the equilibrium time determined at.

1) Modelling of paraquat adsorption isotherms

In this research, the Langmuir, Freundlich and Dubinin-Radushkevitch (D-R) formalisms are shown in Figures 6(a)-(c). These formalisms, which are widely used for the modelling of adsorption isotherms, have been used for the description of paraquat adsorption. The different parameters of Figure 6 obtained by linearization of the equations of the three models are reported in the Table 4. These are the Langmuir (K_L, q_max), Freundlich (K_F, 1/n) and Dubinin-Radushkevitch (K and E) parameters.

q_max (µmol/g), K_L (L/µmol), K_F en (L/g), E (kJ/mol), Langmuir model: C e q e = 1 K L ∗ q max + C e q max , Freundlich model: ln q e = ln K F + ( 1 n ) ln C e , Dubinin - Radushkevich model: ln q e = ln X m − K ′ ε 2 .

➢ Langmuir, Freundlich and Dubinin-Radushkevitch (D-R) model

The comparative study of the regression coefficients (R²) according to Table 4 shows that paraquat adsorption isotherms for most soils are best described by the Freundlich model, followed by the Dubinin-Radushkevich model and finally the Langmuir model (R² Freundlich > R² Dubinin > R² Langmuir).

Two parameters are generated after linearization of the adsorption isotherms from the Langmuir approach: K_L and q_max representing respectively the Langmuir adsorption coefficient and the maximum adsorption quantity of the pesticide on the soil. These parameters were obtained by all soils with the line C_e/q_e = f(C_e) having a positive intercept.

The maximum adsorption capacities (q_max) determined from the Langmuir formalism made it possible to classify the soils according to their performance: 290.697 µmol/g for the NK1 fallow soil, then 235.294 µmol/g for the bare soil of Ebolowa MIN1 and finally 234.192 µmol/g for the NK3 cultivated soil. According to this model, adsorption takes place on homogeneous sites and the process is favourable with the formation of a paraquat monolayer on the surface of the soils at equilibrium.

After linearization of the adsorption isotherms by the Freundlich model, two empirical parameters are generated K_F and 1/n representing respectively the Freundlich adsorption coefficient and the affinity that exists between the pollutant molecules and the adsorption sites of the materials. The lower the value of 1/n, the greater the heterogeneity of the sites [26] . The variability of the 1/n values obtained (0.6 ≤ 1/n ≤ 1) describing the adsorption isotherms of paraquat may reflect different distributions of adsorption sites and thus different processes involved when studying the retention of this pesticide in soil [27] . According to [2] , since 1/n is different from 1, the isotherms are considered as non-linear in the concentration range studied (concentrations introduced at 10⁻⁴ to 10⁻³ mol·L⁻¹ corresponding to 3.5 to 7.0 µmol·L⁻¹). [2] , after a study on the adsorption of terbumeton by soil, shows that this non-linearity has already been observed for this herbicide in a concentration range comparable to that used in this study (0.66 ≤ 1/n ≤ 0.73). According to the same author, two different types of organic matter with different structures and affinities for the pesticide may be responsible for this non-linearity at the soil level. It should be noted that, in soils, certain constituents other than organic matter (mineral constituents) participate in the retention of paraquat. Finally, since the soils studied have very different values of 1/n (Table 4) and the dimension of K_F depends on this value, comparing the values of K_F characterizing paraquat adsorption on different soil samples is not always a requirement.