Oxidized Renewable Materials for the Removal of Cobalt(II) and Copper(II) from Aqueous Solution Using in Batch and Fixed-Bed Column Adsorption

Batch and continuous adsorption of Co and Cu from aqueous solutions by oxidized sugarcane bagasse (SBox) and oxidized cellulose (Cox) were investigated. e oxidation reaction of sugarcane bagasse and cellulose was made with a mixture of H3PO4‒ NaNO2 to obtain SBox and Cox, with the introduction of high number of carboxylic acid functions, 4.5 and 4.8 mmol/g, respectively. e adsorption kinetics of Co and Cu on SBox and Cox were modeled using two models (pseudo-rst-order and pseudo-secondorder) and the rate-limiting step controlling the adsorption was evaluated by Boyd and intraparticle diusion models. e Sips and Langmuir models better tted the isotherms with values of maximum adsorption capacity ( max) of 0.68 and 0.37 mmol/g for Co and 1.20 and 0.57 mmol/g for Cu adsorption on Cox and SBox, respectively. e reuse of both spent adsorbents was evaluated. Adsorption of Cu and Co on SBox in continuous was evaluated using a 2 factorial design with spatial time and initial metal concentration as independent variables and max and eective use of the bed as responses. e breakthrough curves were very well described by the Bohart–Adams original model and the max values for Co and Cu were 0.22 and 0.55 mmol/g. SBox conrmed to be a promising biomaterial for application on a large scale.


Introduction
e pollution of water resources by toxic metal species is one of the main environmental problems in the world. ese inorganic pollutants can cause serious negative e ects on the ecosystem, because they have high toxicity to living organisms and are persistent in the environment due to their nonbiodegradability. Metal ions such as Cu 2+ and Co 2+ are essential, but become toxic by excess exposure. Cobalt is an essential element of vitamin B12, which is necessary for the red blood cell production, but becomes toxic [1] when a high amount is ingested by humans, producing an erythropoietic e ect, cardiomyopathy, hypothyroidism, and polycythemia. On the other band, the immoderate accumulation of copper in the human body can produce hepatic necrosis and Wilson's disease, resulting in abnormalities of the nervous system, liver, kidneys, and cornea [2,3].
Nowadays, several processes have been used to remove inorganic pollutants of di erent sources from water. ese processes are mostly conventional processes based on physicochemical treatments (in combination or not) such as coagulation, occulation, precipitation, and nano-and ultra-ltration [4,5]. However, these processes have a high cost and/or a low e ciency in removing metal ions at low concentrations and, in addition, some of them can generate a large amount of waste that requires to be further disposed of [6]. erefore, these problems have increased the number of studies in this area, aimed at the development of original processes with a higher e ciency and lower cost [7]. In this context, adsorption has been highlighted with the development of new adsorbent materials of lower cost and can be e ciently reused; these materials were obtained from designed syntheses using agroindustrial residues as natural materials [8]. During the two last decades, low-cost bioadsorbents such as lignocellulose biomass [9,10] from agricultural crop waste or chitosan [11][12][13] from shing industry waste have been studied as potential materials to be used in the treatment of e uents containing organic pollutants or toxic metal species. e reuse of agroindustrial residues adds a greater economic value and mitigates a series of environmental problems caused by them. Agroindustrial residues mainly consist of lignocellulose biomass (LB), which can be applied in the adsorption of toxic metal species as an alternative to conventional adsorbents as activated carbons and styrene-divinylbenzene ion-exchange resins, since agroindustrial residues are cheap and e cient metal ion accumulators [14]. However, LB can be chemically modi ed to increase their attraction towards speci c pollutants or to be applied to speci c processes, enhancing their physicochemical properties, such as adsorption capacity, chemical resistance to acidic and basic media, and selectivity [15,16].
Oxidation process is an interesting strategy to modify chemically LB. In addition, the type of oxidizing mixture used will determine the changes in the chemical groups, physical structure, and crystallinity of the LB, producing oxidized materials with di erent physicochemical properties [17]. us, materials with new interesting properties for many purposes can be obtained [18].
Most previous studies using agroindustrial residues for adsorption of toxic metal species are based on batch equilibrium and kinetic studies. Batch adsorption is useful to provide information about the e ciency of an adsorbent and its physicochemical properties; however, batch adsorption may not be the most convenient process to be applied to an industrial scale that requires the treatment of high ow rates. In this way, for a full-scale adsorption process, continuous adsorption is o en preferred [19].
In this study, sugarcane bagasse (SB) and cellulose (Cel) were chemically modi ed by oxidation with a mixture of sodium nitrite (NaNO 2 ) and orthophosphoric acid (H 3 PO 4 ) to obtain oxidized sugarcane bagasse (SBox) and oxidized cellulose (Cox). Adsorption capacity of SBox and Cox for Co 2+ and Cu 2+ ions from spiked aqueous solutions was investigated. e batch adsorption was carried out as a function of the contact time, solution pH and initial Co 2+ or Cu 2+ concentration. e reuse of SBox and Cox was also evaluated. e adsorption of Co 2+ and Cu 2+ in continuous using a xed-bed column lled with SBox was evaluated with a 2 2 experimental design.

Cellulose and Sugarcane Bagasse
Preparation. An analytical mill (Model A11, IKA) was used to mill 50 mm 2 paper sheets of Cel before oxidation reaction. For oxidation reaction, sugarcane bagasse (SB) was treated using the methodology described by Ramos et al. [20]. e SB fraction utilized in this study was that retained on the 0.150 mm (100 mesh) sieve.

Oxidation of Sugarcane Bagasse and Cellulose.
e oxidation reaction of SB and Cel was performed using the procedure previously described by Martins et al. [21]. In a typical methodology, 5.000 g of Cel or SB was weighed into a 250 mL Erlenmeyer ask and then 80.0 mL of H 3 PO 4 (85% wt.%) was added to the ask under agitation at 25°C. A erward, it was added in this ask 4.000 g of NaNO 2 under vigorous agitation for about 10 min and the mixture was allowed to stand for 5 h with the ask opened. en, a vacuum ltration on a sintered glass funnel (porosity 2) was used to separate the reaction mixture from the oxidized material (SBox or Cox), which was washed with distilled water until pH reached neutrality (pH ~ 7). e drying of the oxidized material was performed in an oven at 65°C for 4 h. e oxidized material was kept in a desiccator prior to use.

FTIR Spectroscopy.
To prepare the FTIR KBr pellets of 13 mm, 100 mg of KBr (spectroscopy grade) and 1 mg of sample (dried powder) were mixed and pressed using a hydraulic press (Model 181-1110, Pike Technologies, Canada) at 6 ton for 30 s. e spectrum was obtained from an FTIR spectrometer (Model MB3000, ABB Bomen, Canada) from 500 to 4000 cm −1 with a resolution of 4 cm −1 and 32 scans per sample.

Adsorption Experiments.
e adsorption of Cu 2+ and Co 2+ on SBox or Cox was studied as a function of the contact time, solution pH, and initial Cu 2+ or Co 2+ concentration. e following typical methodology was used in all adsorption experiments, which were carried out in duplicate. Samples of 20 mg of SBox or Cox and 100.0 mL of aqueous Cu 2+ or Co 2+ solution were added to 250 mL Erlenmeyer asks. en, drops of aqueous 0.1 mol/L NaOH or HCl solutions were added to adjust the pH. e asks were placed in an orbital incubator shaker (Model TE-424, Tecnal, Brazil) at 150 rpm and 25°C. At the end of the adsorption experiments, a single ltration ( lter paper JP-41) was carried out to separate the liquid and solid fractions. A ame atomic absorption spectrophotometer (FAAS) (Model SpectrAA 50B, Varian) was used to determine the concentration of Cu 2+ or Co 2+ in the liquid fraction ( Cu = 324.8 nm [22], Co = 240.7 nm [22]). e adsorption capacity, , was calculated utilizing Equation (1): where mmol/g is the amount of Cu 2+ or Co 2+ adsorbed per unit weight of SBox or Cox at time or equilibrium, (L) is the volume of Cu 2+ or Co 2+ solution, (mmol/L) is the initial e adsorption isotherms were obtained to evaluate the in uence of initial Cu 2+ or Co 2+ concentration on metal ion uptake by SBox and Cox by varying the initial Cu 2+ or Co 2+ concentration. e following conditions were employed to obtain the adsorption isotherms: initial Cu 2+ or Co 2+ concentration range from 8 to 100 mg/L, equilibrium time of 3 h and pH 5.5.

Desorption and Reuse of SBox and Cox.
Desorption of SBox and Cox was made to evaluate the possible reuse of these adsorbent materials. e loading of SBox or Cox (50 mg) with Cu 2+ or Co 2+ was performed at pH 5.5 using 100.0 mL of 125 mg/L of Cu 2+ or Co 2+ solution and a time of 3 h. At the end of adsorption experiments, a single ltration was made to recover the loaded adsorbent, which was washed with distilled water to eliminate any nonadsorbed Cu 2+ or Co 2+ . e loaded adsorbent was dried at 80°C in an oven for 1 h. en, 20 mg of Cox or SBox loaded with Cu 2+ or Co 2+ was weighed into Erlenmeyer asks (125 mL) and then 20.0 mL of aqueous 0.5 mol/L HNO 3 solution was added. Desorption experiments were performed in an orbital incubator shaker (Model TE-424, Tecnal, Brazil) at 150 rpm and 25°C for time intervals of 10 and 15 min for SBox and Cox, respectively. e concentration of Cu 2+ or Co 2+ in the desorption solution was determined by FAAS (Section 2.5). e desorption e ciency, des , was calculated utilizing Equation (2), [17]: where mg/L is the equilibrium Cu 2+ or Co 2+ concentration in aqueous HNO 3 solution, (L) is the volume of the aqueous HNO 3 solution, ,max mg/g is the maximum adsorption capacity obtained from loading adsorbent with Cu 2+ or Co 2+ before the desorption experiment and ὔ ads is the weight of SBox or Cox adsorbent contained in SBox or Cox loaded with Cu 2+ or Co 2+ ads, M 2+ [22]. e calculation of ὔ ads is given in Equation (3): ὔ ads / g = ads, M 2+ , max /1000 + 1 .
All reuse studies were made using the following procedure: 20 mg of SBox or Cox recovered from Cu 2+ or Co 2+ desorption was added to Erlenmeyer asks (250 mL) with 100.0 mL of an aqueous 50 mg/L Cu 2+ or Co 2+ solution (pH 5.5). e Erlenmeyer asks were placed in an orbital incubator shaker (Model TE-424, Tecnal) at 150 rpm and 25°C for 3 h. en, a single ltration ( lter paper JP-41) was used to separate the liquid and solid fractions and the equilibrium concentration of Cu 2+ or Co 2+ in the liquid fraction was determined by FAAS (Section 2.5). e re-adsorption e ciency, re-ads , for one adsorption/desorption/re-adsorption cycle was calculated utilizing Equation (4).
where re-ads,max mg/g is the maximum re-adsorption capacity obtained a er one adsorption/desorption/adsorption cycle [22]. re-ads, max is calculated utilizing Equation (5).

Fixed-Bed Column Adsorption Experiments.
e adsorption experiments in continuous were carried out in a jacket glass column (10.0 cm height × 1.0 cm inner diameter) lled with 0.400 g of SBox adsorbent (dry-weight basis). e SBox adsorbent was packed between two layers containing glass beads (10 mm of diameter) and glass wool to give a uniform up ow of the Cu 2+ or Co 2+ solution into the column. For all experiments, the bed height was xed at 3.80 cm. e inlet up ow rate was controlled by a peristaltic pump (Model BP600/2, Milan, Brazil). e temperature of the inlet metal solution and the column was kept at 25 ± 1°C by a thermostatic bath with forced water circulation (Model MA470, Marconi, Brazil). e Cu 2+ or Co 2+ solutions were prepared using a bu er of pH 5.5 (CH 3 COOH/CH 3 COONa). E uent samples were sampled at the top of the column at predetermined time intervals to obtain the breakthrough curve. Metal concentration in the e uent samples was determined by FAAS (Section 2.5). To ensure that metal concentration was in the range of the calibration curve, appropriate dilutions were made. e saturation time and breakthrough point were xed at 95% and 5% of in uent metal concentration, respectively [23].

Design of Experiments.
e continuous monocomponent adsorption of Cu 2+ and Co 2+ on SBox was optimized employing a 2 2 experimental design to evaluate the in uence of the initial Cu 2+ or Co 2+ concentration ( 0 , mmol/L) and spatial time ( , min) on the responses. e value of was calculated using the empty volume of the column (mL) and the ow rate (mL/min). e dependent variables (responses) evaluated were the maximum adsorption capacity ( max , mmol/g) of the bed and the e ective use of the bed ( , cm). Table 1 presents the independent variables (IVs) and their levels used in the continuous adsorption of Cu 2+ and Co 2+ on SBox adsorbent. ey were de ned considering the batch

Batch Equilibrium Data. Adsorption isotherms were
tted by the Freundlich, Langmuir, and Sips isotherm models. e Langmuir [29] isotherm model is expressed by Equation (13).
where max mmol/g is the maximum adsorption capacity of SBox or Cox for a metal ion and (L/mmol) is the Langmuir binding constant. e Freundlich [30] isotherm model is expressed by Equation (14).
where mmol/g (L/mmol) 1/ and are the Freundlich model constants. e parameter is associated to the adsorption intensity. e Langmuir-Freundlich isotherm model, developed by Sips [31], is a hybrid model of the Freundlich and Langmuir isotherm models, which is expressed by Equation (15).
where is a parameter describing the heterogeneity of the system.

Analysis of the Breakthrough Curve.
e area under the breakthrough curves gives the amount of Cu 2+ or Co 2+ adsorbed in the bed [32]. en, the value of max of the bed can be obtained from Equation (16), until the saturation time is reached, as follows: where ̇ is the inlet ow rate (mL/min), is the time, 0 and are the in uent and e uent Cu 2+ or Co 2+ concentrations (mmol/L) and SBox is the weight of SBox (g).
adsorption studies and the lowest ow rate (1.4 mL/min) of the pump. en, 1 min was de ned as the center point of spatial time; and the center point of initial Cu 2+ and Co 2+ concentration was calculated based on the values of obtained from the batch adsorption isotherms. Statistica 10.0 (StatSo , Inc.) routines were used to evaluate the obtained results for analyses of variance (ANOVA), regression coe cients and graphical analysis. To analyze the experimental error, pure error was used and a con dence level equal to 95% was used in the statistical analyses. An IV was statistically signi cant when its -value <0.05.

Modeling the Experimental Data 2.8.1. Batch Kinetic Data.
Pseudo-rst-order (PFO) and pseudo-second-order (PSO) models were used to investigate the adsorption kinetics of Cu 2+ and Co 2+ on SBox and Cox. e PFO kinetic model of Lagergren [24] is expressed by Equation (6).
where 1 min −1 is the pseudo-rst-order rate constant. e PSO kinetic model of Ho and McKay [25] is expressed by Equation (7).
where 2 g/mmol min is the pseudo-second-order rate constant.
In addition, to elucidate the adsorption mechanism of Cu 2+ and Co 2+ on Cox and SBox, two models were used: intraparticle di usion (IPD) and Boyd. e IPD model of Weber and Morris [26] is expressed by Equation (8).
where mmol/g is the intercept and mmol/g min 1/2 is the intraparticle di usion rate constant. e constant may be correlated to the thickness of the boundary layer. e model of Boyd et al. [27] is expressed by Equation (9): where = / is the fractional surface coverage as a function of time and is a function of . Reichenberg [28] provided solutions for Equation (9) using Fourier transform, which are expressed by Equations (10) and (11), as follows: where ( ) is the absolute temperature, is the thermodynamic equilibrium constant (dimensionless) and is the gas constant (8.3144 J/K mol). Liu [39] suggested that (Langmuir constant) can be used to calculate the value of , as given in Equation (23).
where e is the activity coe cient (dimensionless) at equilibrium at 25°C. e value of e can be calculated utilizing the extended Debye Hückel law (Equation (24)), which is applied to ionic strengths up to 0.1 mol/L.
where (mol/L) is the ionic strength, is the charge of the Cu 2+ or Co 2+ and (pm) is the hydrated ion size of Cu 2+ or Co 2+ (600 pm) [40].
e values of were calculated as described by Ramos et al. [20].

Regression Analysis and Error Evaluation. (i) Batch Adsorption Data.
e experimental batch equilibrium and kinetic data were modeled with the PFO and PSO and Freundlich, Langmuir, and Sips models by nonlinear regression (NLR) analysis using Microcal OriginPro®2015. e modeling of the kinetic data with IPD and Boyd models was made by linear regression (LR) analysis of the experimental data. e so ware was set to use the weight method named "statistical" and the Levenberg-Marquardt algorithm. e reduced chi-square 2 red was employed to determine which model best described the experimental adsorption data [41].
(ii) Continuous Adsorption Data. e breakthrough curves obtained were tted using the set of Equations (18)- (20). e algorithm genetic function (ga) of the MATLAB®2010a soware (Mathworks Inc.) was used for adjusting the parameters of the Bohart-Adams model to the continuous experimental data by minimization of the objective function. e objective function used was the root-mean-square error (RMSE) [18], as given in Equation (25).
where is the number of experimental points and is the value of experimental point ,̂ is the predicted value by the model for the experimental point .

Synthesis and Characterization of the Oxidized Materials.
e oxidation reaction of SB and Cel was performed using a mixture of NaNO 2 and H 3 PO 4 to obtain the adsorbent materials SBox and Cox with high amounts of e value of is de ned as the height of the bed that was e ectively utilized in the continuous adsorption process [33]. e value of was obtained from Equation (17), as follows: where is the bed height (cm), 2+ , is the amount of Cu 2+ or Co 2+ adsorbed on the adsorbent until the breakthrough time (mmol) and 2+ , is the amount of Cu 2+ or Co 2+ adsorbed on the adsorbent until the saturation time (mmol). e values of 2+ , and 2+ , were determined by the integration of the area under the breakthrough curves until and , respectively.
(i) Modeling the Breakthrough Curves. e development of mathematical models to predict the breakthrough curve, the mechanism involved in the process and even the evaluation of the e ect of di erent variables on adsorption process is an important demand. However, this is a di cult task for such systems since the concentration pro les in both solid and liquid phases vary with both and [32]. us, analytical models based on modeling the experimental data by means of nonlinear or linear regression analyses with physical meaning are available [34][35][36][37][38]. e original model of Bohart and Adams [38] is successfully employed when the adsorption isotherm is favorable. is model supposes that the rate of adsorption is proportional to the remaining adsorption capacity of the adsorbent and the concentration of the solute, neglecting axial dispersion. Equations (18)- (20) describe the relationship between / 0 and for this model.
where (cm/min) is the interstitial velocity, (mmol/L) is the liquid phase Cu 2+ or Co 2+ concentration initially 0 and at time , (cm) is the column height, (min) is the time, is the column void fraction, (g/mL) is the bulk density and e relation between the adsorbent density and bulk density was used to determine the porosity of the bed, as given in Equation (21).

Calculation of Variation in Standard Free Energy of Adsorption.
e variation in standard free energy of adsorption Δ ads ∘ for batch adsorption was determined by Equation (22) [39].
Advances in Polymer Technology 6 maximum at pH value of 5.5 for all adsorption systems. Interestingly, Cox exhibited a higher value than SBox for both metals studied, with both adsorbent materials even having a similar value for COOH . en, the next studies of adsorption of Cu 2+ and Co 2+ on both materials as a function of the contact time (kinetics) and initial Cu 2+ and Co 2+ concentration (isotherm) were carried out at a pH value of 5.5.
A similar tendency was reported by Gurgel et al. [14,43] and Ramos et al. [20] for the adsorption of Cu 2+ on succinylated cellulose (Cell 6) [14], succinylated sugarcane bagasse (SCB 2) [43] and sugarcane bagasse modi ed with trimellitic anhydride (STA) [20]. SCB 2 has a higher amount of carboxylic acid groups (6.0 mmol/g), but a higher pH PZC value (5.26) [47] than SBox [21]. As a result, SCB 2 only exhibited an adsorptive capacity for Cu 2+ ions from a pH value of 3.0. e pH PZC value of STA is 3.16 because this material possesses carboxylic acid groups according to the optimized procedure reported in a previous study by our research group [21]. e use of this oxidation system (H 3 PO 4 -NaNO 2 ) provides a cheaper methodology to prepare materials containing carboxyl functionality, thus adding value to these materials. e reaction mainly produced the oxidation of the primary R-CH 2 OH groups of SB and Cel into R-COOH groups. e reaction conditions were optimized by response surface methodology (RSM) and design of experiments (DOE).
rough the optimization of the synthesis conditions, SBox and Cox were obtained with a number of R-COOH functions ( COOH ) of 4.5 and 4.8 mmol/g, respectively, which were determined as described by Martins et al. [21]. Cel oxidation was accomplished with a weight gain equal to 7.7%. is can be explained by the fact that the primary alcohol groups were converted into carboxylic acid groups, increasing the molar mass of the biopolymer. A weight loss equal to 18.4% was reported for SB oxidation, because of the presence of oxidant species and the acidic conditions, which degraded lignin and hemicelluloses fractions, increasing the solubility of these fractions in an aqueous medium.
SBox and Cox were characterized by FTIR spectroscopy and solid-state 13 C NMR spectroscopy, as described in our previous study by Martins et al. [21]. e 13 C NMR spectrum of Cox indicated that the main product has carboxylic acid groups but that starting material (nonoxidized cellulose) is also still present in much smaller amounts. e main advantage of these oxidized materials (SBox and Cox) comparing to other carboxylated materials prepared by our research group [14,15,20,22,[42][43][44][45] is that carboxylic acid functions in SBox and Cox are mainly from the oxidation of primary hydroxyl groups (R-CH 2 OH) of cellulose (in SB or Cel). On the contrary, when carboxylic acid groups are introduced in the Cel or SB through esteri cation reaction with cyclic carboxylic acid anhydrides, the application of esteri ed Cel or SB is limited to a pH range where the ester groups may not be easily hydrolyzed in the aqueous phase, e.g., 2-9 [46].  [21]. us, at pH values greater than pH PZC both surfaces of SBox and Cox became predominantly negative due to conversion of carboxylic acid groups into carboxylate anion. erefore, when pH values are higher than 2.7, SBox and Cox have a net negative surface charge, thereby promoting the adsorption of the cationic pollutants, such as Cu 2+ and Co 2+ .

Adsorption
Graphs of q e against initial solution pH for adsorption of Cu 2+ and Co 2+ on Cox and SBox are shown in Figures 1(a) and  1(b), respectively. As expected, the adsorption of both metals increased when the solution pH was higher, reaching a addition, for both adsorbent materials, the values of were higher for Cu 2+ adsorption than Co 2+ adsorption, which will have a great in uence in the behavior of Cu 2+ and Co 2+ adsorption on SBox in a xed-bed column, as presented in Section 3.4.

Adsorption Isotherms of Cu 2+ and Co 2+ on SBox and
Cox. Adsorption isotherms can be employed to describe the di erent a nities of various solutes for the adsorption sites of an adsorbent. To evaluate the equilibrium data, Freundlich, Langmuir, and Sips models, three of various isotherm models available in the literature, were chosen. e isotherm model parameters estimated by tting the models to the experimental data are presented in Table 3 Other important factors to explain the adsorption mechanism are the physicochemical and textural properties of the Cox and SBox adsorbents as well as the mass transport processes. e curves generated from the NLR analyses of the experimental data with the PFO and PSO are presented in Figures 2(a) and 2(b). Table 2 presents kinetic parameters estimated by modeling of the experimental data. Comparing the values of 2 and 2 red presented in Table 2, it was possible to con rm that the PSO model better described the Cu 2+ adsorption by both adsorbent materials. When comparing the values of ,exp of Cu 2+ with those of ,est , it can be observed for both adsorbent materials that the values of ,est estimated by the PSO model were closer to those of ,exp . e values of 2 indicated a higher kinetic rate constant for Cu 2+ adsorption on SBox than Cox. As can be seen in Figures 2(a) and 2(b), the equilibrium time for Cu 2+ adsorption was achieved in 45 and 120 min for SBox and Cox, respectively.
For Co 2+ adsorption on both SBox and Cox, the model that better described the experimental data was the PFO model with the values of ,exp for this model being closer to those values of ,est . e values of k 1 showed a higher kinetic rate constant for Co 2+ adsorption on Cox than SBox. As can be seen in Figures 2(a) and 2(b), the Co 2+ adsorption equilibrium time was achieved in 180 and 120 min for SBox and Cox, respectively. e parameters of IPD and Boyd models estimated by LR analyses of the experimental data are presented in Table 2.
e adsorption process of Cu 2+ and Co 2+ on SBox and Cox exhibited three stages (Figures not shown). ese plots suggest that the adsorption of Cu 2+ and Co 2+ on SBox and Cox was initially governed by the di usion of Cu 2+ or Co 2+ through the thin solvent lm surrounding both Cox and SBox particles and then changed to intraparticle di usion. A erward, equilibrium was reached. Boyd's plots (Figures not shown) were only linear for the initial stage of Cu 2+ or Co 2+ adsorption and did not intersect the origin, suggesting that external mass transfer can be the rate-limiting step governing the initial stage of Cu 2+ or Co 2+ adsorption, which then changed to intraparticle di usion, corroborating the IPD plots [48]. In Advances in Polymer Technology 8 e adsorption isotherms (Figures 3(a) and 3(b)) with initial curvature (type L) suggest that the available active sites are more di cult to nd when the surface coverage increases, indicating that the Cu 2+ and Co 2+ may have been adsorbed onto a monolayer. e evaluation of the values of 2 red and the comparison between the values of max,exp and max,est were used to select the best isotherm model in order to characterize the adsorption systems. Analyzing the values of 2 red , max,exp and max,est presented in Table 3, it can be seen that for Cu 2+ or Co 2+ adsorption on both adsorbents, both the Sips and Langmuir isotherm models can describe very well the experimental data. e Sips model is a hybrid of the Freundlich and Langmuir isotherms. e parameter of the Sips model can be employed to evaluate the degree of heterogeneity of an adsorption system. us, if the value of is closer to unity, this suggests a experimental data of Cu 2+ and Co 2+ adsorption on SBox and Cox are presented in Figures 3(a) and 3(b). e isotherm curves obtained presented a downwardconcave pro le, characterizing a favorable adsorption [35]. e steepest initial slopes of adsorption isotherms of Cu 2+ on SBox and Cox in comparison to Co 2+ adsorption isotherms on SBox and Cox indicate that Cu 2+ ions had more a nity for the adsorption sites of SBox and Cox than Co 2+ ions. is is corroborated by the higher values of for Cu 2+ adsorption than for Co 2+ adsorption (Table 3). In addition, for both Cu 2+ and Co 2+ , even at low metal concentrations, the adsorbed amount was relatively high until reaching a plateau at which the adsorbed amount remained constant. In agreement with the classi cation of isotherms proposed by Giles et al. [49], the pro les of the isotherm curves for all adsorption systems studied corresponded to group L, subgroup 2. From the values of reported in Table 3 for the Sips model, it is concluded that the adsorption systems involving the adsorption of Co 2+ on SBox and Cox are more homogeneous than the adsorption systems involving the adsorption of Cu 2+ on SBox and Cox. It is also noted that the adsorption of Cu 2+ on Cox is signi cantly more heterogeneous than other adsorption systems studied.
Parameter of the Langmuir model is associated to the solute a nity for the adsorbent sites [50]. Comparing the values of of the Langmuir model (Table 3), it is concluded that Cu 2+ has a greater a nity for both Cox and SBox adsorption sites than Co 2+ , as was previously concluded by the initial slopes of Cu 2+ adsorption isotherms in comparison to Co 2+ adsorption isotherms. e solute a nity for the adsorbent sites of the materials increased in the order: Co 2+ -SBox < Cu 2+ -SBox < Co 2+ -Cox < Cu 2+ -Cox. e values of variation in standard free energy of adsorption Δ ads ∘ were negative for all adsorption systems studied, indicating the spontaneous character of the adsorption in the with , which is the time when the value of in the e uent reaches 5% of 0 , to , which is the time when the value of in the e uent reaches 95% of 0 . Figures 5(a) and 5(b) present the breakthrough curves obtained for adsorption of Cu 2+ and Co 2+ on SBox in a xed-bed column. Table 5 presents the values of and max . e results presented in Table 5 show that the higher value of max (0.554 mmol/g) for Cu 2+ adsorption was obtained in experiment 3, while the lower value of max (0.237 mmol/g) was obtained in experiment 2. For Co 2+ adsorption, the higher and lower values of max (0.224 and 0.088 mmol/g) were obtained in experiments 3 and 2, respectively. ese results suggest that higher levels of initial metal concentration and lower levels of spatial time favored higher values of max .
Cox were lower than those for adsorption on SBox, indicating that the COOH taking part on the adsorption of Cu 2+ and Co 2+ by SBox and Cox was di erent. Probably, this was responsible for a decrease in the value of max for SBox in comparison to Cox, as both Cox (4.8 mmol/g) and SBox (4.5 mmol/g) have similar values for COOH .

Desorption and Reuse of SBox and Cox.
Desorption and readsorption were evaluated for both SBox and Cox adsorbents with Cu 2+ and Co 2+ , respectively. e desorption e ciencies des and re-adsorption e ciencies re-ads obtained for both SBox and Cox are presented in Table 4.
It was observed that the desorption of the metals was fast, with a maximum value of des achieved in a short time, i.e., 10 min for SBox and 15 min for Cox. e best result for des was obtained for the SBox-Cu system; however, the other systems showed values of des between 72.3% and 87.7%. Even though SBox and Cox were not completely desorbed, this should not impair their reuse as they still present a large amount of free adsorption sites.
e desorption experiments performed at acidic aqueous medium showed that both metals were fast desorbed, which indicated the ion-exchange was the mechanism controlling both adsorption and desorption of both metals. Xavier et al. [23], who investigated the adsorption/desorption/readsorption of Co 2+ , Cu 2+ and Ni 2+ on trimellitated sugarcane bagasse reported a similar behavior in comparison to that observed in the present study. e infrared spectra of Cox and SBox before and a er desorption of Cu 2+ (Cox-D-Cu 2+ , SBox-D-Cu 2+ ) and Co 2+ (Cox-D-Co 2+ , SBox-D-Co 2+ ), respectively, are presented in Figures 4(a) and 4(b). As can be seen in Figures 4(a) and 4(b), the band at 1730 cm −1 related to C=O stretching in carboxylic acid groups is still present in both spectra of Cox and SBox a er desorption of Cu 2+ and Co 2+ , showing that both adsorbent materials were not degraded a er desorption.
Re-adsorption was performed to evaluate the alternative of reusing both adsorbents. e desorbed SBox and Cox were used in a new adsorption cycle to evaluate their performances as they were not completely desorbed. e obtained results are shown in Table 4. e lower value of re-ads was for Co 2+ -SBox system; however, the other systems showed values of re-ads higher than >98%. erefore, SBox and Cox still exhibited excellent values of re-ads in comparison to their values of ,max showing that these materials are interesting adsorbents from both point of view of chemical resistance to degradation and reuse.

Monocomponent Adsorption of Cu 2+ and Co 2+ on SBox Adsorbent in a Fixed-Bed Column.
e choice of SBox to investigate continuous adsorption of Cu 2+ and Co 2+ in a xedbed column is due to its lower cost in comparison to Cox, although Cox had higher values of max,exp in comparison to SBox. e breakthrough curves plotted as e uent concentration-time pro les provide information about the dynamic behavior of a determined xed-bed column adsorption from the mass transfer zone (MTZ). MTZ is the area of the bed where the adsorption of Cu 2+ and Co 2+ occurs. MTZ varies with increasing ow rate. erefore, it is expected that higher levels of inlet Cu 2+ or Co 2+ concentration and lower levels of spatial time provide a higher concentration gradient di erence between the adsorbed Cu 2+ or Co 2+ at the interface and the Cu 2+ or Co 2+ in the bulk solution, as well as a smaller resistance at the liquid lm, thereby increasing the value of max of the adsorbent.
Higher values of for adsorption of Cu 2+ and Co 2+ on SBox were obtained at the center point for both metal ions, Furthermore, lower levels of initial metal concentration and higher levels of spatial time disfavored higher values of max . In addition, the value of max for Cu 2+ in continuous adsorption was 2.9% lower than in batch adsorption, while the value of max for Co 2+ in continuous adsorption was 65.2% lower than in batch adsorption. e rate-limiting steps of an adsorption process are: (1) the mass transfer of the Cu 2+ or Co 2+ from the bulk phase to the SBox surface; (2) the di usion of the Cu 2+ or Co 2+ into the SBox porous structure; and (3) the adsorption of Cu 2+ or Co 2+ onto the SBox surface [52]. In a mass transfer process, the mass ow is proportional to the concentration gradient di erence.
is is the driving force for the adsorption process. Moreover, the resistance at the liquid lm interface (step 1) decreases (b) F 5: Breakthrough curves for adsorption of (a) Cu 2+ and (b) Co 2+ on SBox adsorbent in a xed-bed column at 25°C (experiments from 1 to 7 refers to the 2 2 experimental design matrices for Cu 2+ and Co 2+ presented in Tables 1 and 5). adsorption, mass transfer of Cu 2+ or Co 2+ from the bulk solution to the SBox surface and porous structure of the SBox as well as with uid ow properties during continuous adsorption into the column [52]. e instantaneous equilibrium produces, under ideal conditions, a shock-like wave which moves as a sharp concentration step through the bed [53], where the entire length of bed is utilized. is suggests that, at the center point, the overall resistance to the adsorption of Cu 2+ or Co 2+ on SBox is reduced, favoring the adsorption. As a consequence, the application of lower levels of 0 and lower levels of τ increased the overall resistance to the adsorption. e 2 2 experimental design was employed to evaluate the e ect of 0 and on the adsorption of Cu 2+ and Co 2+ on SBox in continuous. e analysis of variance (ANOVA) of the experimental data was carried out. e results are shown in Table 6. It is necessary to have a signi cant regression for a welladjusted model, i.e., > tab and < 0.05 together with a non-signi cant lack of t, i.e., < tab and > 0.05. e linear model used to t the response max yielded coe cients of determination 2 of 0.963 for Cu 2+ and 0.857 for Co 2+ . However, the values of 2 for for adsorption of Cu 2+ (0.579) and Co 2+ (0.358) were not high as for max . In addition, the models used to t the DVs max for Co 2+ and for Cu 2+ exhibited a lack of t, indicating that these models cannot predict the behavior of these dependent variables very well.
As can be seen in Table 6, both independent variables were statistically signi cant for response max for Cu 2+ with presenting a negative e ect and 0 a positive e ect on Cu 2+ adsorption. is means that the use of 0 in a higher level and in a lower level favored a higher value of max . e DV max for Co 2+ adsorption was only in uenced by 0 with this variable presenting a statistically signi cant positive e ect. For the DV , both variables were not statistically signi cant for Co 2+ adsorption, and only 0 was statistically signi cant, presenting a positive e ect on Cu 2+ adsorption. e interactions between the independent variables 0 × had no statistically significant e ects on any DVs ( max or ) evaluated for the adsorption of both metals on SBox.
When comparing these results of adsorption in continuous with the pro les of the kinetic curves obtained for batch adsorption of Cu 2+ and Co 2+ on SBox (Figures 2(a) and 2(b)), it can be seen that the spatial time only in uenced Cu 2+ adsorption, which presented kinetic curves with steeper initial slopes than those for Co 2+ adsorption. is suggests that the mass transfer of Cu 2+ or Co 2+ from the bulk phase to the SBox surface may have contributed more to the with 1.67 (experiment 7) and 2.12 cm (experiment 5), respectively. e lower values of , 0.40 cm for Cu 2+ and 0.23 cm for Co 2+ , were obtained in experiment 1. e characteristics of the breakthrough curve are related with the equilibrium   the metals studied were smaller for higher levels of 0 and/or , whereas higher values of B-A were observed for lower levels of 0 and/or . e B-A is a lumped parameter [51], which includes all e ects of internal and external di usion, adsorption kinetics as well as any dispersion in the bed. en, the adsorption in continuous at the conditions evaluated in this study is in uenced by all of these e ects. However, for Cu 2+ adsorption, the B-A value at center point was closer to experiment 1, while for Co 2+ adsorption in the experiment 4 had a value of B-A closer to that obtained at the center point. As the levels of 0 and in experiments 1 and 4 were opposite (Table 5), this indicates that the adsorption of these metals on SBox, in fact, presented di erent rate-limiting steps. 2 and Co 2+ on Di erent Adsorbents.  Immobilized seaweed Sargassum sp. possible to compare the performance of di erent adsorbent materials in continuous from the values of max , 0 , and .

Comparison with Literature Adsorption Data for Cu
Both the values of 0 and a ect the value of max of an adsorbent material. It is possible to have a high value of max using a high value of 0 combined with a low value of or a low value of 0 combined with a high value of . Comparing the performance of SBox adsorbent in continuous with those reported in the literature data, it should be noted that SBox showed a very good performance, mainly for Cu 2+ removal, because it presented a medium value of max , which was obtained from a lower value of combined with a medium value of 0 . However, SBox presented a lower value of max for Co 2+ removal.

Conclusions
e oxidized sugarcane bagasse (SBox) and cellulose (Cox) were successfully synthesized by a cheap chemical modi cation method and shown to be e cient for the removal of Cu 2+ and Co 2+ from spiked single aqueous solutions in both batch and continuous. e adsorption of Cu 2+ on SBox and Cox was better modeled by the pseudo-second-order kinetic model, while the adsorption of Co 2+ on SBox and Cox was better modeled by the pseudo-rst-order kinetic model. e results of equilibrium adsorption were modeled by three isotherm models and the Sips and Langmuir models closely tted the experimental data with maximum adsorption capacities max of 1.20 and 0.57 mmol/g for Cu 2+ and 0.68 and 0.37 mmol/g for Co 2+ adsorption on Cox and SBox, respectively. In addition, Cox and SBox exhibited excellent desorption e ciencies des and re-adsorption e ciencies re-ads , allowing their reuse in further adsorption cycles. SBox was further used in xed-bed column adsorption experiments. e Bohart-Adams original model tted continuous adsorption data very well. e value of max for Cu 2+ (0.55 mmol/g) was closer to batch adsorption, while max for Co 2+ (0.22 mmol/g) was lower than that obtained in batch adsorption. us, the main results obtained in this study show that both adsorbents are interesting materials for a real application in a wastewater treatment plant, mainly SBox, which is a cheaper adsorbent in comparison to Cox.
Data Availability e data used to support the ndings of this study are included within the article.

Conflicts of Interest
e authors declare that they have no con icts of interest.