Efficient removal of Cr (VI) and Co (II) from aqueous solution by activated carbon from Manihot esculenta Crantz agricultural bio-waste

The exponential expansion of industrial activities, caused by high growth of the population and urbanization, has increased industrial wastes and the discharge of a high percentage of toxic heavy metals such as cadmium, mercury, lead, copper, zinc, nickel, chromium and cobalt into the environment, perturbing the environment and posing serious health hazards to humans (Rai et al. 2019). The existence of heavy metals in our aquatic system may be instigated by aqueous solutions of many industries, such as automobile, mining, electroplating, iron–steel and battery industries (Akpomie & Dawodu 2015). Among the heavy metals, cobalt and chromium got more attention due to their dangerous effects. In fact, chromium has many oxidation states and Cr (VI) is the most harmful; its exposure limit is 0.05 mg/L according to the World Health Organization (WHO) (World Health Organization 2003). Chromium can cause cancer in the digestive tract and lungs, epigastric pain, nausea, severe diarrhea, vomiting and haemorrhage (Rangabhashiyam et al. 2016). Cobalt has an occupational exposure limit of 1 μg/L according to WHO (Office of Occupational Health 1996); above this limit it becomes harmful and toxic, inducing breathing problems, such as asthma and pneumonia. A high concentration of cobalt is responsible for vomiting, nausea, vision problems, heart problems and thyroid damage (Farjana et al. 2019).

Hence, various methods and treatment technologies that have been developed for the removal of heavy metals from aqueous solution include precipitation and sludge separation, chemical treatment, oxidation-reduction, ion exchange, reverse osmosis, coagulation, flocculation, electrochemical treatment, membrane processes, evaporation, and adsorption onto activated carbon (Javanbakht et al. 2014). However, these methods have disadvantages, like production of a large quantity of hazardous wastes (Saha et al. 2019), high costs in terms of energy consumption, low performance when heavy metals concentration is below 100 mg/L (Pugazhenthiran et al. 2016). Therefore, there is a necessity to find an effective process to remove heavy metals economically with less cost. Agricultural waste materials were used as activated carbon sources by different researchers. They represent unused resources and also, they are widely available with a great potential to be used as adsorbents (Hashemian et al. 2014).

Cassava (Manihot esculenta Crantz) is a crop largely cultivated in tropical and sub-tropical climates like in Africa (Benin, Nigeria), also in Asia (Indonesia, Thailand) and Latin America (Brazil). Taxonomically it belongs to the Euphorbiaceae family and Manihot genus. Cassava peel is an ideal renewable carbon source and millions of tons of this waste are generated every day (Salgaonkar et al. 2019). Cassava peels are used in many fields like the production of bio-ethanol, compost, also the generation of biofuel, chemicals, and finally in aqueous solution treatment as adsorbent for dyes (Beakou et al. 2017) and heavy metals (Tovar et al. 2019). Nevertheless, to well valorize cassava peel, there is still a need to investigate more the adsorption process to reach the highest metal removal rate. The aim of this work is to display the great potential of cassava peel used as precursor for activated carbon with proper activation. The work includes equilibrium, kinetics and thermodynamic study of adsorption of chromium and cobalt onto cassava peel carbon.

The biomass used in this study, cassava peel tuber bark, was completely washed with distilled water and exposed to the sun light until it was totally dry, then finely ground and stored in a tightly closed container and called Cassava Peel Raw (CPR). The CPR product was ground and sieved to a particle size ranging from 40 to 63 μm. The mixture of the raw material (10 g) and phosphoric acid (mass ratio 1:1) was carried out in a beaker containing 200 mL of solution and heated at 75 °C for 14 hours, and then dried at 100 °C for 24 hours. The obtained product was ground and charred at 350 °C for 2 hours using a furnace model Nabertherm L40/11 Bo Germany under non inert conditions. After cooling in the open air, the support was washed thoroughly with distilled water. The resulting filtrate, poor in phosphoric acid, was dried, ground and soaked in 200 mL of NaHCO₃ (1%) for 14 hours to remove the residual acid and washed with distilled water until neutral pH. Finally, the activated carbon, named Cassava Peel Carbon (CPC), was dried at 60 °C for 6 hours and sieved to sizes between 40 and 63 μm.

Different analytical methods were carried out in order to determine the functional groups, pore distribution, surface morphology and main compositions of the sample. The chemical characterization of the functional groups of CPC was detected using a Fourier Transform Infrared (FTIR) spectrophotometer model Varian 800/Gladiatr (Scimitar Series, Australia/Pike Technologies, USA). Infrared absorbance data were obtained for wavenumbers in the range of 400–4,000 cm⁻¹ using 50 scans collected at 4 cm⁻¹ resolution. Surface area and pore volume calculations were analysed by the Brunauer-Emmett-Teller (BET) and the Barrett-Joyner Halenda (BJH) equations, respectively, with nitrogen adsorption on the activated carbon surface at a temperature of 77 K. The measurements were conducted on Quadrasorb Si (Quantachrome, Germany). All the samples were firstly degassed at 150 °C for 24 hours. X-ray diffraction (XRD) analysis was carried out to identify the crystal structure of CPC by using a powder diffractometer (Phaser D2 Diffractometer, Broker, USA), with a Cu K_α1 radiation (λ = 1.54060 Å) operated at 30 KV and 10 mA. The spectra were recorded with a 2θ angle ranging from 2° to 90°. The surface morphology of CPC was investigated by scanning electron microscopy (SEM) analysis on PHENOM XL (Phenom-World, Netherland), with an accelerating voltage of 10 kV. The composition of CPC was evaluated using Energy-Dispersive X-Ray (EDX) spectroscopy, while the composition of cassava rind carbon was determined by an elementary analyser.

C_i, C_e are the concentrations of metal ions at initial phase and equilibrium (mg/L), respectively. V is the volume of the solution (L), W is the weight of adsorbent (g).

The pH of cobalt and chromium was not modified. The natural pH values of cobalt and chromium solutions were equal to 4.6 and 3.8, respectively. Adsorption kinetic of heavy metals was studied at initial concentration of 20 mg/L of cobalt and chromium ion solutions separately in 200 mL Erlenmeyer flasks, mixed with 0.02 g of activated carbon at a temperature of 25 °C, and a contact time ranging from 5 to 180 min (3 hours). The mixture of metal ion solution and the adsorbent was shaken at 300 rpm in a double-jacketed beaker. Adsorption isotherms were also performed, using various chromium and cobalt concentrations in the range of 10–40 mg/L, with adsorbent dose of 0.02 g. The latter mixture was agitated in a rotary shaker at 300 rpm, in which temperature was maintained at 25 °C with a thermostat (Jp Selecta Frigiderm, Spain) for 3 hours. The thermodynamic adsorption was evaluated at different temperatures of 15, 25, and 35 °C; while keeping other parameters fixed: (i) activated carbon dose of 0.02 g, (ii) solution concentration of 35 mg/L, and (iii) magnetically stirring the mixture at 300 rpm in a double-jacketed beaker for 3 hours.

The fitting of adsorption kinetics, adsorption isotherms and the thermodynamic adsorption were performed separately with OriginPro data analysis and graphing software version 2018.

To understand more the mechanism of cobalt and chromium adsorption onto CPC, desorption experiments were performed. After performing the equilibrium study with different initial heavy metal concentrations ranging from 10 to 25 mg/L, the activated carbon particles were collected by filtration and then dried at 60 °C for 2 hours. Then, 100 mL of NaOH, distilled water and HNO₃ solutions were used separately for the regeneration study.

The X-ray diffraction measurements were performed to evaluate the crystallographic structure of CPR and CPC. The spectra in Figure 1 show the XRD patterns for both materials. It is noted that the presence of very sharp diffraction peaks and the absence of broad peaks reveal an increasing regularity of crystalline structure and a better layer alignment for CPC, while CPR exhibited an amorphous structure with one sharp diffraction peak. Furthermore, the main diffraction peak was around 2θ = 26°, corresponding to the (002) plane, a characteristic peak of graphite (Zbair et al. 2018). The calcination temperature (T = 350 °C) suggests the start of the graphitization process and the formation of a partial nanocrystalline structure of the compound (Mirgorod et al. 2016).

As expected, the cassava peel biomass after activation shows a higher development of surface area and pore volume. Indeed, the dehydration process by a chemical activating agent like H₃PO₄ increases the surface area and porosity (Heidarinejad et al. 2020). The specific surface area of CPC given by the BET model was found 618.265 m²/g and the volume of pores is 1.53610⁻¹cm³/g with an average pore diameter of 3.016 nm. According to the IUPAC classification (Gaya et al. 2015), CPC showed mostly the development of a mesoporous structure. In addition, the shape of the N₂ adsorption–desorption isotherm was type IV (a), showing the presence of well-developed microporosity. Therefore, the adsorption capability of CPC is related to a particular disposition of both micro- and mesoporous texture. N₂ adsorption–desorption isotherm (supplementary figure 10), pore size distribution (supplementary figure 11) and incremental pore volume of CPC (supplementary figure 12) are given in the Supplementary material.

The morphology and structure of the cassava peel carbon CPC were observed by SEM. As shown in Figure 2, the surface area of CPC is predominantly that of an amorphous compound with crystalline components. Moreover, the existence of some particulate matter on the external surface of CPC might be related to the activating agent H₃PO₄.

Table 1 gives the composition on a mass basis of CPR from elemental analysis and the composition, also on a mass basis, of CPC evaluated by EDX, which gives a semi quantitative analysis (just carbon and nitrogen). As noticed, the chemical composition of carbon did not significantly change after the activation step (from 37.7% to 38.65%), while there was a decrease in the oxygen level (from 55.9% to 42.72%. Maybe this is related to the fact that the calcination did not take place in an inert atmosphere. EDX also showed the presence of 14.09% phosphorus. This is due to the use of phosphoric acid as an activating agent, which allows us to predict the acidic surface nature of our adsorbent.

From Figure 3, FTIR analysis of CPR and CPC exhibited a profound transformation during activation, where the disappearance of the absorption band at 1,367 cm⁻¹ in the CPC spectrum (present in the CPR spectrum) was noted. This band is attributed to the C-H symmetrical bending of the hemicellulose methoxyl group (Boukir et al. 2019). Additionally, the disappearance of the most intense peak in the CPR spectrum at 995 cm⁻¹, associated with the C-O bond of cellulose from C-O-C groups (Ilyas et al. 2019), was detected in the CPC spectrum. In the CPC spectrum, the peaks located at 770 cm⁻¹, 667 cm⁻¹ and 581 cm⁻¹ corresponded to high crystalline cellulose with the forms I_α and I_β (Lopes et al. 2014). Furthermore, the peak at 1,147 cm⁻¹ found in both spectra belonged to the aromatic C-H of lignin (Lupoi et al. 2015). The peak at 1,575 cm⁻¹ is attributed either to the C = C vibration of the aromatic nucleus (Kubovský et al. 2020) or to the C = O stretching vibration of the lignin. FTIR results confirmed the disappearance of hemicellulose, crystallization of a part of the cellulose, and better appearance of lignin during the transformation of CPR into CPC. The disappearance of hemicellulose is favoured by the calcination temperature used, and the appearance of crystalline cellulose is in agreement with the XRD analysis. The functional groups of lignin and crystalline cellulose are therefore those involved in the adsorption of heavy metal ions.

The effect of pH was investigated in preliminary tests. The two metals exhibited better adsorption for pH between 3 and 6. Under those acidic conditions, the CPC surface is protonated by H⁺ ions, which favours the electrostatic attraction between and the positively charged surface while creating a competition between the H⁺ ions and Co²⁺ for the adsorption sites. As a result, when the pH increases, the adsorption capacity of chromium and cobalt decreases and increases, respectively (chromium is better adsorbed at pH = 3, while the cobalt adsorption is favoured at pH = 6). However, the adsorption capacities of chromium and cobalt at these optimum pHs are close to the natural pH of each metal solution. For this reason, the kinetic, equilibrium and thermodynamic studies were performed without pH modification.

The variation between final and initial pH was also plotted versus the initial pH to evaluate the pHpzc (Figure 4). The pHpzc is a vital parameter of adsorbents to explore the adsorption mechanisms. The value at which pH final was equal to pH initial was accepted to be pHpzc, and the charge surface is neutral at this pH value; the value of pHpzc was found to be 4.59 for CPC, indicating that the surface of the CPC is negatively charged at pH higher than 4.59 and positively charged at pH lower than 4.59. Moreover, under pHpzc, the adsorption of each metal was favoured.

The influence of initial concentration on the adsorption of chromium and cobalt is shown in Figure 5; the removal of both heavy metal ions decreased with an increase in the initial concentration at a fixed adsorbent dosage (0.02 g) and a temperature of 25°. The reason was that at higher concentrations, the ratio of the initial number of available adsorption sites to cobalt and chromium molecules was low, the number of available adsorption sites became lower, and thus the removal of the two heavy metals was reduced.

Figure 6 exhibits the influence of temperature on cobalt and chromium adsorption. It is presented that there is an increase from 234.16 mg/g to 342.5 mg/g in the adsorption capacity of cobalt as temperature increases from 15 °C to 55 °C, and an increase from 152.20 mg/g to 266.86 mg/g in the adsorption capacity of chromium as temperature increases from 15 °C to 55 °C. The increase of the equilibrium adsorption with increase in temperature indicated that this process was endothermic; this can be explained by the increase in the mobility of cobalt and chromium ions caused by the decrease of the medium's viscosity with increasing temperature of the solution.

where Q_t (mg/g) is the amount of metal ion on the surface of the sorbent at time t (min), Q_e (mg/g) is the amount of metal ion adsorbed at equilibrium. k₁ (min⁻¹) and k₂ (g/mg·min) are the dynamics constants. To determine the goodness of fit of the two kinetic models, the Reduced Chi-Square χ² and R-Square R² were used as error parameters for each model. The results are shown in Table 2.

The kinetic profiles for the adsorption of the two heavy metal ions are presented separately in Figure 7 and indicate that for both heavy metal ions (chromium and cobalt), the adsorption took place via two stages. The first stage is a fast sorption uptake that occurred within 40 minutes of metal ion and activated carbon contact. The second stage is a slow phase of metal ion removal that developed from 50 min until 180 min when a quasi-stabilised state was presumed to have been reached. The adsorption capacity values at the equilibrium after 180 min were 182.77 mg/g for cobalt ion and 158.47 mg/g for chromium ion. The two-stage kinetic profiles for cobalt and chromium ions can be associated with the nature of available surface sites for adsorption. When the contact between cassava peel carbon and the heavy metal ion is established at the onset of sorption, there are a good number of available sites for sorption to occur; hence the fast metal ion uptake observed. However, as the uptake proceeds and available sites are occupied, the rate of further adsorption decreases due to the repulsive forces between the already adsorbed metal ion and the incoming sorbate and the limitation of available sites for occupation.

Table 2 shows that the pseudo-second order model gives the best fit for the two heavy metal ions, according to: (i) the R² coefficient values (0.997 and 0.986 for cobalt and chromium, respectively) higher than the pseudo-first order model (0.973 for cobalt and 0.945 for chromium), (ii) the Reduced Chi-Square (0.068 for cobalt and 0.21 for chromium) lower than the pseudo first order model (0.62 and 0.91 for cobalt and chromium, respectively); similar results was obtained by Scheufele et al. (2019), who studied the biosorption of direct black dye by cassava root husks. The value of Reduced Chi-Square (χ²) of cobalt indicates that the adsorption of this heavy metal onto CPC is best described by the pseudo-second order model. Also, the pseudo-second order constant k₂ for cobalt was 0.00187 g/mg·min, while that for chromium was 0.00254 g/mg·min. This indicates that the sorption of chromium ions onto the CPC was faster than that of cobalt ions; the same result was discussed by Swelam et al. (2018), who also investigated the removal of cobalt ions from aqueous solution using raw and modified rice straw. Since the activated carbon has both micropores and mesopores, the adsorption of the ion with the bigger radius (cobalt) may be favoured. Moreover, (iii) the calculated adsorption capacity values at the equilibrium (186.75 mg/g and 155.99 mg/g for cobalt and chromium, respectively) for the pseudo-second order model are closer to the experimental values (182.77 mg/g and 158.47 mg/g for cobalt and chromium, respectively).

Based on these results, the pseudo-second order is the best kinetic model of cobalt and chromium adsorption onto the CPC.

If R_L > 1, adsorption is unfavourable; if R_L = 1, adsorption is linear; if R_L < 1, adsorption is favourable; and if R_L = 0, adsorption is irreversible.

From Table 3, it can be observed that the Freundlich constant K_F, which relates to adsorption capacity, was 129.81 (mg⁽¹⁻ⁿ⁾ Lⁿ/g) for chromium and 153.79 (mg⁽¹⁻ⁿ⁾ Lⁿ/g) for cobalt. This indicates that the loading of cobalt was higher than that of chromium on the CPC adsorbent. Furthermore, the Freundlich constant n, which relates to the intensity of adsorption, was 3.66 for chromium and 8.92 for cobalt, indicating more favourability of the cobalt ion adsorption on CPC adsorbent. The Langmuir constant K_L, which characterizes the energy constant, was 6.59 (L/mg) for chromium and 1.038 (L/mg) for cobalt, indicating that the chromium adsorbed onto the surface of cassava peel carbon has a lower heat of adsorption. The maximum adsorption capacity Q_max for cobalt was 301.63 mg/g and that of chromium was 166.35 mg/g. This indicates that cobalt adsorption was higher on the adsorbent than that of chromium ion. The Redlich-Peterson constant β value was 0.97 for chromium and 0.89 for cobalt, both values tending to 1, so the Redlich-Peterson model reduces to Langmuir model. Furthermore, the R_L value was 0.011 and 0.076 for chromium and cobalt, respectively, indicating that the adsorption is favourable.

To determine the goodness of fit of the Langmuir, Freundlich and Redlich-Peterson isotherm models to the experimental data, non-linear regression and error parameters were used and defined for each model (coefficient of determination (R²) and the Reduced Chi-Square) (Scheufele et al. 2019). The comparison of the fitting parameters of these three models indicates the best model that describes well the isotherm adsorption. Table 3 indicates that the Redlich-Peterson model has high R² values for both heavy metals (0.999 for chromium and 0.993 for cobalt) compared with the Langmuir model (0.998 for chromium and 0.989 for cobalt) and the Freundlich model (0.98 for chromium and 0.977 for cobalt). Moreover, the Reduced Chi-Square of the Langmuir model has lower values (0.054 for chromium and 0.21 for cobalt) compared with the Redlich-Peterson model (0.53 for chromium and 0.86 for cobalt) and the Freundlich model (0.838 for chromium and 0.25 for cobalt). Finally, we can conclude that the Redlich-Peterson isotherm is the better, but β values tend to 1, which means that the Langmuir model can explain the adsorption mechanism and the general tendency of the two heavy metal ions is to fill the first layer of the surface of the CPC. We can conclude a monolayer adsorption nature.

In addition, Co (II) adsorption capacity is higher than Cr (VI), while chromium adsorption kinetics is faster. The particular disposition of the microporous and mesoporous pore observed in textural properties may affect the adsorption mechanism. In fact, the difference between the ionic radius of Co (II) (79 ppm) and Cr (VI) (58 ppm) can make the adsorption of chromium faster, as it can fit in micropores (Arshadi et al. 2014). But as the activated carbon is mostly mesoporous, the amount of cobalt adsorbed is higher.

From Table 4 we can notice that the mango kernel activated carbon has been found to be less effective as adsorbent material for chromium, with an adsorption capacity of 7.8 mg/g. Date press cake activated carbon has shown the highest adsorption capacity for chromium at 282.8 mg/g. Also in the same table, we can observe that potato peel activated carbon has shown the highest adsorption capacity for cobalt at 405 mg/g compared with other material such as lemon peel activated carbon, rice straw AC, hazelnut shell AC and CPC, which showed only 22.00, 28.5, 13.88 and 301.63 mg/g, respectively.

where K_d is the equilibrium constant (L/g), R (8.134 J/(mol·K)) is the gas constant, and T(K) is the absolute temperature. ΔH_ads and ΔS_ads were obtained from the slope and intercept of the Van't Hoff's plot of ln(K_d) versus 1/T.

The outcomes are presented in Table 5. The negative value of the Gibbs free energy (ΔG_ads) explains the spontaneous nature of the adsorption of cobalt and chromium ions onto CPC; indeed, the adsorption process becomes more spontaneous with rising temperature, which makes the adsorption process more favourable. The endothermic nature of the adsorption process of these two heavy metals is confirmed by the positive values of enthalpy ΔH_ads. In addition, generally the enthalpy ΔH_ads values less than 40 kJ/mol·K suggest that the adsorption is physical adsorption and reversible. This endothermic character also supports what we explained above regarding the increase in adsorption capacity of CPC with increasing temperature. Furthermore, the standard entropy ΔS_ads has positive values and indicates the increase of randomness at the solid-solution interface during adsorption. Finally, as mentioned above, CPC had sufficient surface area, a certain number of surface functional groups, and the largest adsorption capacities for cobalt and chromium. And the adsorption process showed better spontaneity.

The desorption efficiencies for heavy metal ions by using three kinds of desorbing agent (HNO₃, distilled water, NaOH) were in the range of 96.83%–88.28% for cobalt and 96.72%–87.09% for chromium (Figure 9). We can notice also from the figure that desorption in the acidic media (HNO₃) for the two heavy metal ions was more rapid and presented higher cobalt and chromium recovery from the CPC than in basic and neutral media. In acidic medium, the functional groups in the CPC become protonated and do not attract the positively charged metal ions, and so the protons (H⁺) replace the bound metal ions. Therefore, the metal ions cannot compete with H⁺ ions for adsorption sites, and subsequently metal ions are released from the adsorbent surface into the solution.

In conclusion, cassava peel carbon (CPC) has shown its potential to be an effective agricultural waste material valorized for heavy metal removal from aqueous solution. It is found that the kinetic adsorption data of cobalt and chromium was well fitted to the pseudo-second order model, while the adsorption isotherms were fitted to the Langmuir isotherm model, which confirm the monolayer adsorption nature of cobalt ions and chromium ions on CPC. Likewise, the thermodynamic study showed that the adsorption process on the CPC is endothermic, spontaneous and follows a physisorption process. Differences between Cr (VI) and Co (II) adsorption performances may be attributed to the pH effect. Electrostatic attraction/repulsion explains the fact that cobalt adsorption capacity was higher than chromium.

All relevant data are included in the paper or its Supplementary Information.