Environmental surface chemistries and adsorption behaviors of metal cations (Fe3+, Fe2+, Ca2+ and Zn2+) on manganese dioxide-modified green biochar

The facile preparation and modification of low-cost/efficient adsorbents or biochar (CP) derived from the carbonization of palm kernel cake (lignocellulosic residue) has been studied for the selective adsorption of various metal cations, such as Fe3+, Fe2+, Ca2+ and Zn2+, from aqueous solution. The CP surface was modified with KMnO4 (CPMn) and HNO3 (CPHNO3) in order to improve the adsorption efficiency. The physicochemical properties of the as-prepared adsorbents were investigated via BET, pHpzc, FT-IR, Boehm titration, TG-DTG, XRD and SEM-EDS techniques. The surfaces of all adsorbents clearly demonstrated negative charge (pHpzc > pH of the mixture solution), resulting in a high adsorption capacity for each metal cation. Fe2+ was found to be more easily adsorbed on modified CP than the other kinds of metal cations. Synergistic effects between the carboxylic groups and MnO2 on the surface of CPMn resulted in better performance for metal cation adsorption than was shown by CPHNO3. The maximum adsorption capacities for Fe3+, Fe2+, Ca2+ and Zn2+ using CPMn, which were obtained from a monolayer adsorption process via Langmuir isotherms (R2 > 0.99), were 70.67, 68.60, 5.06 and 22.38 mg g−1, respectively. The adsorption behavior and monolayer-physisorption behavior, via a rapid adsorption process as well as single-step intra-particle diffusion, were also verified and supported using Dubinin–Radushkevich, Redlich–Peterson and Toth isotherms, a pseudo-second-order kinetic model and the Weber–Morris model. Moreover, the thermodynamic results indicated that the adsorption process of metal cations onto the CPMn surface was endothermic and spontaneous in nature. This research is expected to provide a green way for the production of low-cost/efficient adsorbents and to help gain an understanding of the adsorption behavior/process for the selective removal of metal ions from wastewater pollution.


Introduction
The existence of toxic/heavy metals in water is regarded as a part of the inorganic pollution that has become a worldwide issue, due to their deleterious effects on the environment, including human life and ecological systems. 1,2 Recently, Fe, Zn and Ca have raised concern and been identied as toxic metals that are generally discovered in ground/surface water, leading to many problems in industrial processes, such as slag formation in boilers and tube failure. In fact, the concentrations of Fe and Zn, as reported by the World Health Organization (WHO), should not exceed over 0.3 and 0.01 mg L À1 , respectively. The three water treatment or purication steps in industrial processes are: (I) the removal of suspended solids from the system via precipitation; (II) the removal of organic substances using bacterial degradation; and (III) the removal of inorganic substances including various metals. 1,[4][5][6] Several conventional technologies such as coagulation, co-precipitation, ltration membranes and ion-exchange have been mainly identied as good processes for the removal of metal ions from wastewater. 3,7,8 However, the disadvantages of these technologies include the high costs of equipment, the large amount of expensive chemicals required and the complications of largescale operations. Adsorption processes have been proven to be an effective and reliable method for the removal of metal ions from wastewater. 9 Activated carbon is accepted as an excellent adsorbent since it presents a high surface area, large porosity and long-term stability. The development and modication of carbon surfaces results in improved adsorption rates and capacities. 10 Commercial activated carbon is chiey used in practical wastewater treatment processes. However, it involves higher production costs than charcoal, for instance, as the pyrolysis process is performed at high temperatures of $900-1000 C under a N 2 atmosphere using hardwoods as the feedstocks. 11 Recently, it has been found that the performance of commercial activated carbon is solely limited to the excellent adsorption of non-polar or slightly polar molecules, such as reactive dyes, iodine, and organic compounds (lipids and phenols); however, the adsorption capacities for polar molecules, including cations and anions such as Ca 2+ , Mg 2+ , Cr 3+ , Cd 2+ , Fe 2+ , Fe 3+ , Zn 2+ , AsO 4 3À , Cr 2 O 7 2À and MnO 4 À , are quite low. 12 For cation adsorption mechanisms, the existence of oxygen/nitrogen functional groups is required, such as carboxylic, hydroxyl and amine groups with lone pair electrons, which could serve as Lewis bases on the surface of activated carbon for the attraction of cations (Lewis acids) via electrostatic forces with the generation of co-ordinate covalent bonds. As mentioned above, it is still necessary to develop and improve the selective functional groups on the surface of activated carbon to improve the adsorption capacity and selectivity for cations. To date, many researchers have been searching for effective waste biomass to use for the production of low-cost and highly efficient activated carbons. In our previous work, we studied the production of activated carbon from waste lignocellulosic biomass, such as Terminalia catappa seeds and Cerbera odollam seeds, and the results indicated that these activated carbon sources exhibited better adsorption capacities for Ca 2+ , Fe 2+ , Fe 3+ and Cr 3+ than commercial activated carbon. It is also considered that the production costs are much cheaper, compared with commercial activated carbon. 12 However, due to the existence of too small amounts of these kinds of waste biomass sources, in a further idea we tried to apply charcoal sold in a local market as an adsorbent feedstock. The advantages of charcoal or biochar are: (I) the production costs are low; (II) the production process is not complicated; and (III) villagers can produce by themselves great amounts of this adsorbent. It should be mentioned here that 1 kg of charcoal or biochar produced from Leucaena leucocephala (Lam.) de Wit costs 20 Baht; if this is applied to activated carbon production, 0.9 kg of product can be obtained. For the modication process, Liu et al. 11 prepared FeCl 3 -modi-ed activated carbon (positive charge/Lewis acid) using an impregnation process. They found that FeCl 3 could serve as an oxidizing agent to increase the amount of carbonyl groups on the carbon surface, leading to the enhancement of the adsorption ability towards Cr 2  The maximum Cd 2+ adsorption capacities of the activated carbon before and aer KMnO 4 modication were 116 and 217 mg g À1 , respectively. They also reported that KMnO 4 was indicated to be the best oxidizing agent and expected that it could be applied to the adsorption of other metal ions. However, for the preparation processes above, it needs to be calcined at very high temperatures, which requires a huge amount of energy. Moreover, complex technology is required. Thus, developing an environmentally friendly modication process with the use of simple technology, as well as controlled production costs, is necessary. 15 Also, the adsorption capacities and uptake abilities should be improved from current values. Based on the above discussion, the abundant existence of carbonyl groups on the surface of biochar produced from a carbonization process in a conned space needs further modication through the functionalization of carboxylic groups on the surface for the better adsorption of cations. It should be noted that a harsh oxidizing agent is required due to the specic structure of charcoal. Considering economic factors, the objectives of this work are to improve and develop the performance of low-cost biochar for the selective adsorption of multiple metals with high capacities. Here, palm kernel cake (biomass residue derived from the palm oil production process) was applied as a feedstock for biochar-adsorbent production. The surface modication of the as-prepared biochar was investigated using oxidizing agents such as HNO 3 and KMnO 4 at different ratios. The adsorption ability of each prepared adsorbent was compared with commercial activated carbon through considering the adsorption capacities for metal cations and iodine number values. The optimum conditions, such as the pH value, ion concentration and adsorbent amount, and the reusability were investigated in detail. The physical and chemical properties of the as-prepared adsorbents, such as BET surface area, functional group composition, acidity/basicity, morphology, thermal decomposition properties and pH pzc , were studied. To obtain more details regarding the adsorption behavior, various parameters, such as the surface chemistry, isotherms, kinetic models, and intra-particle diffusion and thermodynamic data, were determined systematically. To the best of our knowledge, there are no reports in the literature on the use of this kind of low-cost adsorbent for the selective removal of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ along with studies of its respective adsorption behavior. It is expected that the developed low-cost adsorbent could be applied to metal removal from wastewater from industrial processes.

Materials and reagents
Palm kernel cake was obtained from the local palm oil industry in Thailand, and it was carbonized at 350 C under conned conditions until no smoke appeared during the carbonization process. The carbonized carbon samples obtained from the above-mentioned method are denoted as CP. Here, CP was crushed and sieved through a 400 mesh sieve before any activation or modication processes. Commercial activated carbon (ACC) was purchased from Sigma-Aldrich. All chemical reagents used were of analytical grade and purchased from Merck and Sigma-Aldrich. Stock solutions (1000 mg L À1 ) of Fe 2+ , Fe 3+ , Ca 2+ and Zn 2+ were prepared via dissolving FeSO 4 $7H 2 O, FeCl 3 -$6H 2 O, CaCl 2 and ZnCl 2 in deionized water, respectively.

Adsorbent preparation and modication
In brief, the adsorbent was prepared and modied as follows: (I) Physical activation: CP was heated at 400-600 C for 1-5 h. Here, it should be mentioned that, for instance, CP activated at 500 C for 1 h was denoted as CP501.
(II) Chemical activation (HNO 3 ): 5 g of CP was mixed with 50 mL of conc. HNO 3 and reuxed at 100 C for 4 h. The samples obtained from this process are denoted as CPHNO 3 .
(III) Chemical activation (KMnO 4 ): 5 g of CP was mixed with 50 mL of 0.02 mol L À1 KMnO 4 and stirred at ambient temperature for 4 h. The samples obtained from this process are denoted as CPMn.
(IV) Chemical activation (HNO 3 + KMnO 4 ): 5 g of CP was mixed with 50 mL of 0.02 mol L À1 KMnO 4 and stirred at ambient temperature for 4 h. The samples obtained from this process are denoted as CPMix.
Details of the adsorbent characterization methods carried out (e.g., BET, pH pzc , FT-IR, Boehm titration, TG-DTG, XRD and SEM-EDS studies) are provided in the ESI. † 16

Adsorption studies
All the adsorbents prepared using the above-mentioned method were primarily investigated to nd the optimum conditions for the adsorption of Fe 2+ , Fe 3+ , Ca 2+ and Zn 2+ before further studies of isotherms, kinetic models and thermodynamic adsorption. The metal cation adsorption procedure was as follows: 0.1 g of adsorbent was added to 25 mL of 250 mg L À1 Fe 2+ , 250 mg L À1 Fe 3+ , 25 mg L À1 Ca 2+ or 100 mg L À1 Zn 2+ and stirred at a speed of 150 rpm at a temperature of 303.15 K for 30 min. Meanwhile, studies of the adsorption of 0.05 mol L À1 I 2 were also carried out under the same conditions as the metal cation adsorption process. Aer nishing the processes, the adsorbents were separated via ltration and the obtained solutions were then analyzed to nd the remaining concentrations of Fe 2+ , Fe 3+ , Ca 2+ , Zn 2+ and I 2 . The concentrations of Fe 2+ and Fe 3+ were analyzed using a 1,10-phenanthroline method with a UV-visible spectrophotometer at a wavelength of 510 nm (Genesys 20), while the Zn 2+ concentration was analyzed using a ame atomic absorption spectrometer (Thermo Scientic iCE 3000). The concentrations of Zn 2+ and I 2 were determined via titration with EDTA and S 2 O 3 2À solutions, respectively. The amount of adsorption at equilibrium (q e , mg g À1 ) was calculated according to eqn (1): where C e is the equilibrium concentration of the adsorbate (mg L À1 ), C 0 is the initial concentration of adsorbate (mg L À1 ), V is the volume of solution (L), and M is the mass of adsorbent (g).

Investigation of adsorbent reusability
Before reusability testing, used adsorbent (CPMn-Fe 2+ , Fe 3+ , Ca 2+ or Zn 2+ ) was obtained based on the optimum desorption process mentioned above. In short, 0.1 g of regenerated adsorbent (CPMn) was added to 25 mL of 280 mg L À1 Fe 2+ , 210 mg L À1 Fe 3+ , 30 mg L À1 Ca 2+ or 75 mg L À1 Zn 2+ and stirred at a speed of 150 rpm at a temperature of 303.15 K for 30 min.

Investigation of adsorption isotherms, kinetics and thermodynamics
The details relating to adsorption isotherms, kinetics and thermodynamics are provided in the ESI, † Fig

Adsorption ability and behavior
Fig. S1 † shows the burn loss percentage results of biochar at different temperatures. As expected, increasing the calcination temperature or time obviously promoted the burn loss of the biochar. The results of I 2 and Fe 2+ adsorption using the various adsorbents prepared at different temperatures are shown in Fig. S2. † It is found that ACC provided maximum I 2 adsorption of 1435 mg g À1 , suggesting that the surface area, porosity and non-polarity of the surface of ACC was much higher when compared with CP. In contrast, the lowest Fe 2+ adsorption of 8.56 mg g À1 was also found for ACC, while all types of CP exhibited better Fe 2+ adsorption capacities. This phenomenon could be attributed to the existence of high levels of carbon with low oxygen amounts on the ACC surface, resulting in the excellent adsorption of non-polar I 2 molecules. 6 In the case of the adsorption behavior of Fe 2+ , CP had high levels of oxygencontaining functional groups, such as carbonyl groups with lone electron pairs (Lewis base), leading to the good adsorption of Fe 2+ cations (Lewis acid) via attraction through electrostatic forces with the generation of co-ordinate covalent bonds. It should be noted that increasing the calcination temperature and time resulted in a slight reduction in Fe 2+ adsorption capacity, while the I 2 adsorption capacity was increased to some extent; this was probably due to the decomposition of carbonyl groups on the adsorbent surface into CO 2 , leading to a decrease in the polarity of the adsorbent structure. Thus, CP without physical activation was selected for further study. Fig. 1 shows the results of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption using various adsorbents, including those modied with KMnO 4 and HNO 3 .
Here, CP could be directly modied with KMnO 4 and Fe 3+ without pyrolysis activation, which could greatly reduce the production costs of adsorbents. Interestingly, one can clearly see that CPMn exhibited the highest adsorption performance towards Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ , followed by CPHNO 3 . This should result from the effects of modication with KMnO 4 and HNO 3 through which the functional groups on the CP structure were changed into carbonyl/carboxylic groups as per the following equations: Considering the cation sizes (Ca 2+ (231 pm) > Fe 2+ (76 pm) > Zn 2+ (74 pm) > Fe 3+ (64 pm)) and E 0 values (Ca 2+ (À2.869 V) < Zn 2+ (À0.76 V) < Fe 2+ (À0.44 V) < Fe 3+ (À0.04 V)), the Ca 2+ cation was found to be most difficult to adsorb when compared with the other cations. Also, CP and ACC without any modication could not adsorb Ca 2+ from aqueous solution. This suggests that the carboxylic groups on CP modied with KMnO 4 and HNO 3 and MnO 2 formation on the surface of CP occurring from the reduction of KMnO 4 were required for the Ca 2+ adsorption mechanism, as well as for the adsorption of other cations. Here, comparing the highest adsorption capacities of all types of studied cations, Fe 2+ was found to be much more easily adsorbed with CPMn (60.18 mg g À1 ). It should be mentioned here that even though the modication of CP with HNO 3 was successfully achieved, lower adsorption performance towards various metal cations was still found compared with CP modi-ed with KMnO 4 . Moreover, it requires a large amount of water for the washing procedure and pH adjustment. Therefore, we proposed to use KMnO 4 for adsorbent modication, which presented low-production costs and a non-complicated process. In order to get more detail, the effects of various ratios of CP to KMnO 4 or HNO 3 on I 2 and Fe 2+ adsorption are shown in Fig. S3. † It is found that the I 2 adsorption capacity was increased to some extent upon only increasing the ratio of CP to HNO 3 . Highest I 2 adsorption capacities of 866.62 mg g À1 and 689.71 mg g À1 were found for CPHNO 3 1 : 50 and CPMn 1 : 10, respectively. Also, CPMn 1 : 10 provided a maximum Fe 2+ adsorption capacity. Increasing the KMnO 4 or HNO 3 amount had no signicant effect on improving the Fe 2+ adsorption efficiency. Here, KMnO 4 was presented to be a better oxidizing agent than HNO 3 , probably due to the formation of MnO 2 crystals, which promoted the adsorption capabilities for metal cations. Based on the above results, since CPMn had better performance for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption, CPMn with an optimum ratio of CP to KMnO 4 (1 : 10) was selected for further studies, such as investigating the effects of pH, desorption and reusability, as well as equilibrium adsorption. Table 1 shows the surface chemical properties of CP, CPHNO 3 and CPMn. As obtained, the pH pzc values of all adsorbents had higher values than the pH of the mixed solution (adsorbent + each cation: Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ ), indicating that the surfaces of all adsorbents were occupied by negative charge (pH < pH pzc ), resulting in the high Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption abilities. The acidity and basicity levels of CP, CPHNO 3 and CPMn are shown in Table  1. It is found that CP modied with HNO 3 (CPHNO 3 ) and KMnO 4 (CPMn) had more carboxylic groups than CP without modication. It should be noted that the basicity of the adsorbent was obviously increased in the case of CPMn. This phenomenon could be generally attributed to the formation of a brown precipitate of MnO 2 , which could react with HCl leading to an increase in the adsorbent basicity. Moreover, from checking the results of the physical properties of each adsorbent, CPHNO 3 and CPMn had higher surface areas than CP (Table 1). This might be one reason for the improved Fig. 1 A comparison of the Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption efficiencies using various adsorbents before and after modification with KMnO 4 and HNO 3 . adsorption capacities for metal cations. A highest surface area of 56 m 2 g À1 was found for CPMn, which had a surface area more than nine times that of CP. This should result from the MnO 2 contribution, which was obtained from the modi-cation of CP by KMnO 4 via an in situ reduction process. FT-IR spectra of CP calcined at different temperatures and times are shown in Fig. S4. † As observed, CP calcined at 400-600 C for 1-4 h presented peaks at wavenumbers of 3500 cm À1 (-OH and -NH), 2900 cm À1 (C-H), 1735 cm À1 , 1600 cm À1 (C]O and C]C), 1217 cm À1 , 1114 cm À1 and 1030 cm À1 (C-O), which should be attributed to the vibrations of various acidbase functional groups, such as carboxylic, lactone, phenolic and amino groups. 17 Here, the calcination temperature and time had signicant effects on the enhancement of carbonyl group levels. However, the decomposition of these functional groups on activated CP may occurred with conversion into CO 2 in the case of excessively high calcination temperatures and long calcination times being applied, leading to deactivation and the low performance of the adsorbent for metal cation removal. In addition, C-O vibrations at a wavenumber of 1030 cm À1 were found for CP calcined at 600 C for 4 h (CP604). This could be explained as resulting from the decomposition of CP into ash upon exceeding the temperature for the combustion process. Interestingly, as shown in Fig. S5, † the wavenumbers of the oxygen functional group peaks from the CPHNO 3 structure at 1735 cm À1 (C]O), 1550 cm À1 (-COO-), 1250 cm À1 (-C-O) and 3450 cm À1 (-OH) strongly increased and presented higher vibrational intensities than CP, CPMn and CPMix, which could be assigned to the high amount of carboxylic groups on the surface (Rios et al., 2013). 18 Moreover, spectra from CP modied with HNO 3 and KMnO 4 at different ratios of CP to HNO 3 or KMnO 4 are also shown in Fig. 2. An increase in the ratio of CP to HNO 3 from 1 : 10 to 1 : 50 and ratio of CP to KMnO 4 from 1 : 10 to 1 : 40 resulted in obvious increases in the carbonyl group peaks at wavenumbers of 1730 and 1600 cm À1 , respectively, on the surface of modied CP. Here, the optimum ratios of CP to HNO 3 and KMnO 4 were 1 : 50 and 1 : 40, respectively. In order to conrm the existence of MnO 2 on the CP surface aer modication with KMnO 4 together with an in situ reduction process, the XRD pattern of CPMn, as shown in Fig. 3, clearly exhibits diffraction peaks from MnO 2 at 2theta values of 29 (310), 38 (211) and 40 (301), indicating that MnO 2 was obviously formed via KMnO 4 reduction. 14 Fig. 4 shows TG-DTG proles with the thermal decomposition data from various adsorbents. Moisture/water evaporation from the structures was found for all adsorbents at a thermal decomposition temperature of about 100 C, while the carboxylic groups decomposed at 200-650 C. 18 In the case of CPHNO 3 , the thermal decomposition peaks of the carboxylic groups clearly appeared at 200-400 C and 400-650 C. It should be mentioned here that a higher decomposition temperature of 450-650 C was found for CPMn, which might be attributed to the decomposition of carbonyl groups occurring from oxidation by KMnO 4 . The morphologies of the adsorbents CP, CPMn and CPMn aer the adsorption of metal cations, such as Fe 3+ , were observed via SEM (Fig. 5). As seen, the surface morphology of CP was smooth, whereas that of CPMn was rough and uneven (curved petal-like walls). Aer the cation metal adsorption  process, no signicant change in the surface morphology of CPMn was found in this study. Also, the presence of metal species was also observed via SEM-EDS (Fig. S6 †). No bulk MnO 2 particles were found on the modied CP, suggesting that the MnO 2 particles were of very small size and were well dispersed on the adsorbent surface. Moreover, the existence of metal species such as Mn, Fe, Ca and Zn was also conrmed aer the adsorption process was carried out.   mentioned here that at a low pH value, H 3 O + might react with MnO 2 (basic) existing on CP, resulting in a reduction of the adsorption capacity towards metal cations. In other words, as the surface negative charge density of CPMn reduced with a decrease in the pH value, electrostatic repulsion between the positively charged metal cations and the surface of the adsorbents was also inated, which may result in a decrease in the adsorption capacity. Moreover, it is also found that when the pH value in solution was >7, the adsorption capacities for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ continuously increased. This suggests that these ions possibly could react with OH À to form Fe(OH) 3 , Fe(OH) 2 , Ca(OH) 2 and Zn(OH) 2 precipitates at pH > 7. Thus, the high Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption abilities at pH > 7 did not result from using CPMn. Here, the adsorption capacities for each metal cation in this study were quite different, probably due to the different effects of ion size, stability and other factors. Based on these results, it is not necessary to adjust the pH value of ion solutions when using CPMn as an adsorbent for further studies. Fig. 7 shows the desorption efficiencies of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ using CPMn. Increases in the desorption percentages of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ were graphically observed when the volume of 1.0 mol L À1 HNO 3 was increased, suggesting that the presence of HNO 3 in high amounts could demolish the force of attraction between functional groups with lone pairs of electrons (Lewis base) and the four metal cations (Lewis acids). This indicates that spent adsorbent could be easily regenerated upon washing with HNO 3 solution. However, it is also possible that the simultaneous desorption of metal cations and anchored MnO 2 from the adsorbent surface occurs in the case of using high concentrations of HNO 3 . In the case of pure water and 1.0 mol L À1 NaCl, the results are not shown here since no desorption of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ from CPMn was signicantly detected, even though the amounts of pure water and 1.0 mol L À1 NaCl were adjusted. This indicates that the attraction forces between the metal cations and CPMn were quite strong. Therefore, HNO 3 was chosen as a suitable chemical for application to Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ desorption.

Reusability of the adsorbents
In this study, spent CPMn was regenerated by using 1.0 mol L À1 HNO 3 for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ desorption, and it was reused again under the same conditions (Fig. 8). It is found that aer regenerated CPMn was applied to Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption in a second cycle, the adsorption capacity for each metal cation decreased by about 45% when compared with the fresh sample. Also, this can be compared with regenerated CP without KMnO 4 modication, where only an 18% reduction was found during reusability testing. This phenomenon could be   explained based on the reaction of HNO 3 with MnO 2 on CP during the regeneration process. From these results, it can be seen that CPMn is not suitable for reuse. Moreover, during the general regeneration of spent adsorbent, strong acid and a lot of water is required for the washing procedure, leading to an increase in the production costs of the adsorbent. However, considering the production costs of biochar from palm kernel cake or other agricultural waste biomass, the normal prices of biochar/carbonization (0.6 USD per kg), grinding (0.6 USD per kg) and KMnO 4 chemical activation (4.0 USD) and the overall cost (6.3 USD per kg) were found to be much cheaper than that of commercial activated carbon (127 USD per kg). Meanwhile, CPMn also exhibited better performance than commercial activated carbon for metal cation adsorption. Thus, it is not necessary to regenerate/reuse it via washing with acid solution.

Adsorption equilibrium
In this study, the adsorption behavior was studied; six mathematical isotherm models, the Langmuir, Freundlich, Temkin, Dubinin-Radushkevich, Redlich-Peterson and Toth models, were applied based on varying concentrations of adsorbate ion solution, as shown in Fig. 9. For non-linear methods, a trial/ error process, which was reasonable to apply via computer operations, was applied to assign the isotherm, kinetic, intraparticle diffusion and thermodynamic parameters to maximize the correlation coefficients obtained from experimental results. In this study, R 2 was used to test the best-tting isotherm using the equation expressed in eqn (4): 19 where q m (mg g À1 ) is the equilibrium capacity obtained from the isotherm model that was calculated using the Excel solver, q e (mg g À1 ) is the equilibrium capacity obtained from experimental data, and q e (mg g À1 ) is the average of q e . The details of each isotherm model and the applied equations are also provided in the ESI. † As shown in Fig. 9, metal adsorption capacities using CPMn were in the order: Fe 2+ > Fe 3+ > Zn 2+ > Ca 2+ . Table 2 shows the equilibrium isotherms calculated via non-linear methods for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption using CPMn. Considering the R 2 values, the adsorption results for all metal cations using CPMn were well tted using the Langmuir isotherm model (R 2 > 0.99), and the ts were also found to be greater than those obtained using the Freundlich adsorption isotherm, indicating monolayer adsorption of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ (Lewis acid) on the CPMn surface via chemical attraction through electrostatic force with the generation of co-ordinate covalent bonds. The q max values of Fe 2+ , Fe 3+ , Zn 2+ and Ca 2+ adsorption using CPMn were 70.67, 68.60, 22.38, and 5.06 mg g À1 , respectively, which also exhibited the order: Fe 2+ > Fe 3+ > Zn 2+ > Ca 2+ . For the Temkin isotherm model, one can see that only the R 2 values for Fe 3+ and Zn 2+ adsorption were close to 1. It should be mentioned here that Zn 2+ was the most strongly anchored on the surface of CPMn, as veried by it having the highest A value (182.83 L mol À1 ) calculated from the Temkin model. The Dubinin-Radushkevich parameters could be considered as useful since suitable R 2 values (>0.9) were found for all metal cations. Here, the E values for the adsorption of the metal cations Fe 3+ , Zn 3+ and Ca 2+ were lower than 8 kJ mol À1 , corresponding to a physisorption process. 24 In the case of Fe 2+ , the E value was in the range of 8-16 kJ mol À1 , which could be attributed to chemisorption behavior between Fe 2+ and the CPMn surface. Meanwhile, the q s value calculated from the Dubinin-Radushkevich isotherm was close to the q max value calculated from the Langmuir isotherm, indicating the accuracy of the applied model. Also, the highest value of q s using CPMn was found for Fe 2+ adsorption. This result is in good agreement with those mentioned above. For the Redlich-Peterson and Toth models, they were applied to support the assumptions of the Langmuir isotherm model through considering the g constant. As expected, the g constant, Th and R 2 values of these models were close to 1, conrming a monolayer adsorption process, especially for CPMn-Fe 2+ , CPMn-Ca 2+ and CPMn-Zn 2+ . In addition, it is clearly found that a maximum q e N value was obtained for Fe 3+ adsorption, indicating the potential shown by CPMn for metal cation removal in this study. Fig. 10 show the adsorption capacities of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ aer different contact times using CPMn. Here, rapid initial adsorption was shown aer a contact time of 3 min, with the same trend for all adsorbed cations; then adsorption gradually increased and became close to equilibrium within 40 min. This phenomenon might be explained based on the existence of many adsorption sites with adequate vacant sites present during the inception phase. In addition, the absorption site saturation point of CPMn for each metal cation was easily reached as the time extended. To obtain more details regarding the adsorption behavior, as shown in Table 3, the Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption mechanisms on CPMn were studied and found to be controlled by pseudo second-order kinetics, based on the greater R 2 coefficient (R 2 > 0.99) compared to pseudo rst-order kinetics, indicating a very fast adsorption process. Meanwhile, the calculated q e values were also very close to the q e values derived from the experimental results, suggesting that this model has high accuracy. In addition, the experimental observations were further investigated via intraparticle diffusion using the Weber-Morris model, as given by eqn (5), 26 and the results are shown in Fig. 11.

Adsorption kinetics and intra-particle diffusion
where k p (mg g À1 min 0.5 ), the intraparticle diffusion rate constant, is obtained from the slope of the straight-line plot of q t versus t 0.5 . It should be noted that from this model, the inner diffusion was controlled by the mass transfer rate if a plot of q t vs. t 0.5 was linear and passed through the origin. In the case of a plot that did not pass through the origin, a lm diffusion adsorption mechanism as well as a chemical reaction would be the ratecontrolling steps. As observed in Fig. 11, one can see that no multi-linearity was found in these plots, which could be described through a single-step occurring for the Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption mechanisms. In other words, the straight lines did not pass through the origin as mentioned above, indicating that the adsorption behavior of all metal cations in the aqueous phase involved a complex process. From Table 2 Langmuir, Freundlich, Temkin and Dubinin-Radushkevich isotherm models and their parameters with correlation coefficients obtained using non-linear methods. The initial concentration of adsorbate (C 0 ): Fe 3+ ¼ 210-260 mg L À1 , Fe 2+ ¼ 200-250 mg L À1 , Ca 2+ ¼ 20-30 mg L À1 , Zn 2+ ¼ 65-90 mg L À1 ; 25 mL; adsorbent amount ¼ 0.10 g; time (t) ¼ 60 min; temperature (T) ¼ 303.15 K; pressure (P) ¼ 101 kPa. All parameters were calculated via nonlinear regression in the Excel program these results, it could be concluded that the rapid intra-particle diffusion of metal cations onto CPMn occurred only via onestep adsorption, probably due to interactions between the active sites and cations, the high availability of free sites on external surfaces and/or the presence of easily accessible sites. 27 3.8. Adsorption thermodynamics Fig. 12 shows a comparison of the adsorption abilities for each metal cation at different temperatures using CPMn. One can obviously see that the increasing the adsorption temperature from 303.15 to 328.15 K resulted in the enhancement of the adsorption capacities for all metal cations, especially Fe 2+ and Fe 3+ , indicating an endothermic adsorption process. This phenomenon should also be ascribed to the fact that a rise in temperature could enhance and promote the diffusion rates of metal cations across the external boundary layer and in the internal pores of the CPMn structure, as well as reduce the viscosity of the solution. To evaluate the effects of temperature on Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorbed on CPMn, the thermodynamic parameters, namely standard enthalpy (DH, kJ mol À1 ) (DH phys ( DH chem , typically DH chem > 100 kJ mol À1 ), standard entropy (DS, J mol À1 K À1 ) and Gibbs standard free energy (DG, kJ mol À1 ), were investigated. 28 As shown in Table 4, the negative values of DG obtained from Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption at different temperatures using CPMn were in a range between À0.71 and À5.36 kJ mol À1 , suggesting a physisorption process. 29 It should be mentioned here that the change in free energy for physisorption is in a range between À20 and 0 kJ mol À1 , while for chemisorption it is in a range between À80 and À400 kJ mol À1 . Here, the DG value obviously decreased to some extent with an increase in the adsorption temperature. This change in free energy could be attributed to the spontaneous natures of the Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption processes using CPMn. The endothermic nature of Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption was revealed, while positive DH and DS values indicated irreversible and Table 3 Pseudo-first-order and pseudo-second-order kinetic model parameters. The initial concentration of adsorbate (C 0 ): Fe 3+ ¼ 210 mg L À1 , Fe 2+ ¼ 280 mg L À1 , Ca 2+ ¼ 20 mg L À1 , Zn 2+ ¼ 75 mg L À1 ; 25 mL; adsorbent amount ¼ 0.10 g; time (t) ¼ 3-70 min; temperature (T) ¼ 303. randomness processes. 30,39 A physisorption process was clearly found based on DH < 100 kJ mol À1 , which was in good agreement with the results from the Dubinin-Radushkevich isotherm model. For comparison the adsorption performance is compared with the previous literature, as shown in Table 5, with higher adsorption capacities based on q e max than other previously developed adsorbents. This indicates that not only was a low-cost adsorbent easily prepared, but it also presented better capacity for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption. This phenomenon should be attributed to the contributions of Mn and biochar. From these results, CPMn could be considered as a promising low-cost/efficient adsorbent for the selective removal of metal cations from aqueous solution.

Conclusions
The preparation and modication of CP (low-cost/highly efficient adsorbent) derived from palm kernel cake was successfully achieved for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorption. CP was found to favor the adsorption of polar Fe 2+ molecules, while non-polar I 2 molecules were excellently adsorbed on ACC, as a result of the different polarities of the adsorbent surfaces. CPMn exhibited better performance for metal cation adsorption than CPHNO 3 , suggesting contributions from the carboxylic groups and MnO 2 on the surface, which was conrmed from the TGA, FT-IR and XRD results. Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ adsorbed on CPMn could be easily desorbed using 1.0 mol L À1 HNO 3 solution for a regeneration process. A pH range of about 6 to 7 was appropriate for the adsorption of all metal cations. The adsorption capacities for Fe 3+ , Fe 2+ , Ca 2+ and Zn 2+ using CPMn were 52.39, 60.15, 4.48 and 16.27 mg g À1 , respectively, while CP without any modication showed capacities of 19.3, 36.02, 0.05 and 6.57 mg g À1 , respectively. For the adsorption behaviors towards various metal cations using CPMn, several models were calculated with non-linear forms, such as the Langmuir, Dubinin-Radushkevich, Redlich-Peterson and Toth isotherm models, as well as a pseudo-second-order kinetic model; these were tted based on R 2 values close to 1, suggesting monolayer physisorption with a rapid adsorption process. The single-step diffusion in this study was veried using the Weber-Morris model. Moreover, the adsorption process was found to be spontaneous and endothermic in nature, as described via thermodynamic investigations. This research is expected to show that CPMn could be really applied to the practical process of wastewater treatment.