Cu(II) and As(V) Adsorption Kinetic Characteristic of the Multifunctional Amino Groups in Chitosan

: Amino groups in the chitosan polymer play as a functional group for the removal of cations and anions depending on the degree of protonation, which is determined by the solution pH. A hydrogel beadlike porous adsorbent was used to investigate the functions and adsorption mechanism of the amino groups by removal of Cu(II) as a cation and As(V) as an anion for a single and mixed solution. The uptakes of Cu(II) and As(V) were 5.2 and 5.6 µ mol / g for the single solution and 5.9 and 3.6 µ mol / g for the mixed solution, respectively. The increased total capacity in the presence of both the cation and anion indicated that the amino group (NH 2 or NH 3 + ) species was directly associated for adsorption. The application of a pseudo second-order (PSO) kinetic model was more suitable and resulted in an accurate correlation coe ﬃ cient (R 2 ) compared with the pseudo ﬁrst-order (PFO) kinetic model for all experimental conditions. Due to poor linearization of the PFO reaction model, we attempted to divide it into two sections to improve the accuracy. Regardless of the model equation, the order of the rate constant was in the order of As(V)-single > Cu(II)-single > As(V)-mixed > Cu(II)-mixed. Also, the corresponding single solution and As(V) showed a higher adsorption rate. According to intraparticle and ﬁlm di ﬀ usion applications displaying two linear lines and none passing through zero, the rate controlling step in the chitosan hydrogel bead was determined by both intraparticle and ﬁlm di ﬀ usion.


Introduction
The application of natural biopolymers, chitosan, has been reported for water and wastewater treatment as adsorbents owing to their natural abundance, nontoxicity, hydrophilicity, and biodegradability [1,2]. Chitosan has been utilized in various ways according to their required use, such as (nano)particles [3], beads [1], or membranes [4] in solution, and have demonstrated high effective removal efficiencies and uptakes for cations and anions. Similar to alginate, chitosan bead was formed by covalent crosslinking [5] and H-bonds interaction [6]. Chitosan exhibits the ability to remove cation heavy metals [7]. Moreover, chitosan is useful for removing anions, such as As(V), phosphate [8], dyes [9], and toluene [10]. This is because of the unique characteristic linked to the amino group. The removal by the amino group can be explained by Equation (1) for cations and Equation (2) for anions. The amino functional groups exist as NH 3 + and NH 2 via protonation and deprotonation, respectively, and these bind with an anion via electrostatic forces and with a cation heavy metal via surface complexation [11].

Adsorption Reaction Model
The experimental data were analyzed by using the following four kinetic models.

Pseudo First-Order Model
The assumption of pseudo first-order (PFO) is that (1) the rate of occupation of sorption sites is proportional to the number of unoccupied sites and (2) sorption only occurs on localized sites and there is no interaction between the sorbed ions, which correspond to the monolayer of adsorbates on the adsorbent surface [29,30]. The equation of PFO, which was first proposed by Lagergreen (1907), can be represented as follows [31]:

Pseudo Second-Order Model
The pseudo second-order (PSO) model was first described by  for the kinetic process of the adsorption of divalent metal ions onto peat [32]. While the PFO model estimated the initial stage of adsorption, the adsorption behavior was described during the entire range or final stage of adsorption process via the PSO model [18,33]. The following PSO Equation (5) can be presented: dq t dt = K 2 (q e − q t ) 2 (5)

Intraparticle Diffusion
The intraparticle diffusion model, which considers pore diffusion, was developed and proposed by Weber and Morris (1963) as follows [25]: The internal diffusion model assumes that internal diffusion of the adsorbate is the slowest step, resulting in the rate-controlling step during the adsorption process, and the adsorption is instantaneous [33].

External Mass Transfer
To identify the external mass transfer, it is assumed that the diffusion of the adsorbate is controlled by the liquid film around the adsorbent. Spahn and Schlunder (1975) suggested the following equation [26]:

Batch Test
A series of batch tests was conducted for the single and mixed solutions of Cu(II) and As(V) at pH~5.5, and the uptakes are shown in Figure 1. The individual uptakes were~5.2-5.6 µmol/g for Cu(II) and As(V) for the single solution, indicating the difference of two uptakes is less than 10%. For the mixed solution, the uptakes for Cu(II) and As(V) were 3.6 (30% decrease) and 5.9 (5% increase) µmol/g, respectively, resulting in a total capacity increase to 9.4 µmol/g, which reached 87% of the sum of each ion. This increased total capacity of chitosan can be explained by the hybrid characteristic of the amino group in chitosan. The available sorption sites are separated for Cu(II) and As(V) based on Equations (1) and (2), respectively. As a result, Cu(II) and As(V) can independently interact with NH 2 and NH 3 + , respectively. However, Figure 2, calculated based on the Katchalsky equation [34], shows that over 80% of amino groups are protonated as NH 3 + at a pK a of 6.0-7.0, which enhances the adsorption for As(V) via electrostatic forces. Theoretically, when the pK a (depending on the molecular weight and the degree of deacetylation) is estimated to be 5.5 for the inner chitosan polymer, the reaction is proportionally related to the presence of amino groups. Another assumption is additional association with Cu(II) except for the amino groups. Domard (1987) suggested that the oxygen on the OH group in the chitosan polymer also contributes to chelation with heavy cation metals [35].

Removal Efficiency
The adsorption rates of Cu(II) and As(V) in the form of removal efficiency (%) vs. time (h) for single and mixed solutions are shown in Figure 3. The dash lines indicate the point of 90% uptake. For all cases, it is likely that the adsorption rate was initially faster and then became slower. This phenomenon may be governed by the chance of collision between ions and the adsorbent. At the beginning of the reaction, the higher concentration in the experiment led to surface adsorption with the higher driving force, which reduced the mass transfer resistance in the bulk and film layer [21]. Then, 90% of the sorption removal for Cu(II) was reached at 6 and 24 h, and for As(V), at 0.5 and 6 h for the single and mixed solutions, respectively. The sorption equilibrium for As(V) was more quickly achieved than for Cu(II), and the single solution was at least four times faster than the mixed solution. The relative long equilibrium of Cu(II) is similar to that in Kannamba et al. (2010) [36], who reported that 12 h is required for equilibrium using chitosan flakes. As(V) reached equilibrium at 5 h using 0.25-0.35 mm-sized chitosan [37].

Removal Efficiency
The adsorption rates of Cu(II) and As(V) in the form of removal efficiency (%) vs. time (h) for single and mixed solutions are shown in Figure 3. The dash lines indicate the point of 90% uptake. For all cases, it is likely that the adsorption rate was initially faster and then became slower. This phenomenon may be governed by the chance of collision between ions and the adsorbent. At the beginning of the reaction, the higher concentration in the experiment led to surface adsorption with the higher driving force, which reduced the mass transfer resistance in the bulk and film layer [21]. Then, 90% of the sorption removal for Cu(II) was reached at 6 and 24 h, and for As(V), at 0.5 and 6 h for the single and mixed solutions, respectively. The sorption equilibrium for As(V) was more quickly achieved than for Cu(II), and the single solution was at least four times faster than the mixed solution. The relative long equilibrium of Cu(II) is similar to that in Kannamba et al. (2010) [36], who reported that 12 h is required for equilibrium using chitosan flakes. As(V) reached equilibrium at 5 h using 0.25-0.35 mm-sized chitosan [37].
A longer equilibrium time was required for the mixed solution regardless of the species. This was because of the competition derived from the increased mass of the adsorbate for the limited

Removal Efficiency
The adsorption rates of Cu(II) and As(V) in the form of removal efficiency (%) vs. time (h) for single and mixed solutions are shown in Figure 3. The dash lines indicate the point of 90% uptake. For all cases, it is likely that the adsorption rate was initially faster and then became slower. This phenomenon may be governed by the chance of collision between ions and the adsorbent. At the beginning of the reaction, the higher concentration in the experiment led to surface adsorption with the higher driving force, which reduced the mass transfer resistance in the bulk and film layer [21]. Then, 90% of the sorption removal for Cu(II) was reached at 6 and 24 h, and for As(V), at 0.5 and 6 h for the single and mixed solutions, respectively. The sorption equilibrium for As(V) was more quickly achieved than for Cu(II), and the single solution was at least four times faster than the mixed solution. The relative long equilibrium of Cu(II) is similar to that in Kannamba et al. (2010) [36], who reported that 12 h is required for equilibrium using chitosan flakes. As(V) reached equilibrium at 5 h using 0.25-0.35 mm-sized chitosan [37].

Nonlinear Pseudo First-Order (PFO) and Second-Order (PSO) Models
From Equation (4) for PFO, the integrating form for the boundary conditions of t = 0 to t = t and for qt = 0 to qe = qt is expressed as nonlinear (Equation (8)).
After separating the variables from Equation (5) for PSO and by applying the boundary condition (t = 0 to t = t and qt = 0 to qe = qt), the nonlinear PSO follows Equation (9).
Nonlinear PFO (NPFO) and nonlinear PSO (NPSO) were modeled using Equations (8) and (9), and the calculated variables, including qecal., K, and R 2 , are listed in Figure 4 and Table 1. Upon comparing the R 2 , although both NPFO and NPSO showed good agreement (>0.94), except for Cu(II)mixed, NPSO achieved higher accuracy for all experimental conditions. Additionally, the qecal. was closer to qeexp. for NPSO than for NPFO. As shown in Figure 4, the simulated NPFO line did not likely represent a sudden change in qe at 3-6 h leading to a lower R 2 . Moreover, the pseudo n order equation, Equation (10), based on adsorption capacity was simulated to determine the accurate kinetic order (n) [39]. The calculated n values were 1.98, 2.01, 1.84, and 1.98 for Cu(II)-single, As(V)single, Cu(II)-mixed, and As(V)-mixed, respectively.
Although methodological bias is introduced for PSO [40], the NPSO is the more appropriate kinetic model with a higher applicability for PSO, indicating that the adsorption process is mainly A longer equilibrium time was required for the mixed solution regardless of the species. This was because of the competition derived from the increased mass of the adsorbate for the limited absorbable sites despite their separation. According to prior studies [38], increasing the initial concentration retards the equilibrium despite the higher concentration gradient.

Nonlinear Pseudo First-Order (PFO) and Second-Order (PSO) Models
From Equation (4) for PFO, the integrating form for the boundary conditions of t = 0 to t = t and for q t = 0 to q e = q t is expressed as nonlinear (Equation (8)).
After separating the variables from Equation (5) for PSO and by applying the boundary condition (t = 0 to t = t and q t = 0 to q e = q t ), the nonlinear PSO follows Equation (9).
Nonlinear PFO (NPFO) and nonlinear PSO (NPSO) were modeled using Equations (8) and (9), and the calculated variables, including q e cal., K, and R 2 , are listed in Figure 4 and Table 1. Upon comparing the R 2 , although both NPFO and NPSO showed good agreement (>0.94), except for Cu(II)-mixed, NPSO achieved higher accuracy for all experimental conditions. Additionally, the q e cal. was closer to q e exp. for NPSO than for NPFO. As shown in Figure 4, the simulated NPFO line did not likely represent a sudden change in q e at 3-6 h leading to a lower R 2 . Moreover, the pseudo n order equation, Equation (10), based on adsorption capacity was simulated to determine the accurate kinetic order (n) [39]. The calculated n values were 1.98, 2.01, 1.84, and 1.98 for Cu(II)-single, As(V)-single, Cu(II)-mixed, and As(V)-mixed, respectively.
Although methodological bias is introduced for PSO [40], the NPSO is the more appropriate kinetic model with a higher applicability for PSO, indicating that the adsorption process is mainly associated with chemisorption [18,41], whereas PFO proceeds by diffusion through a boundary [22]. The rate constant K was considered to determine the adsorption equilibrium [42]. Both absolute K1 and K2 are in the order of As(V)-single > Cu(II)-single > As(V)-mixed > Cu(II)-mixed. This observation shows that a higher K was calculated for the single solution and As(V), regardless of the kinetic model. Table 1. Kinetic parameters using nonlinear PFO and PSO models for Cu(II) and As(V) in single and mixed solutions.   The rate constant K was considered to determine the adsorption equilibrium [42]. Both absolute K 1 and K 2 are in the order of As(V)-single > Cu(II)-single > As(V)-mixed > Cu(II)-mixed. This observation shows that a higher K was calculated for the single solution and As(V), regardless of the kinetic model.

Linear PFO and PSO
Linearization of PFO and PSO was attempted to determine the best-fitting model by minimizing the error distribution between experimental and predicted values. Linear PFO (LPFO) and PSO (LPSO) were derived from Equations (8) and (9), respectively, and expressed by Equations (11) and (12), respectively. ln(q e − q t ) = ln q e − K 1 t (11) Equations (11) and (12) are frequently applied to fit experiment data instead of nonlinear PFO and PSO models. The calculated parameters are listed in Table 2 and shown in Figures 5 and 6 for LPFO and LPSO, respectively. Note that Equation (12) is one of the linearized PSO expressions among the five equations derived from Equation (9) and is the most valid and widely used expression, showing the highest R 2 and similarity between q e exp. and q e cal. [43,44]. Our study also obtained the most valid R 2 from Equation (12) (data not shown). showing the highest R 2 and similarity between qeexp. and qecal. [43,44]. Our study also obtained the most valid R 2 from Equation (12) (data not shown).   Table 2 and Figures 5 and 6 clearly verify the effect of linearization. While the fitting accuracy by R 2 increased from 0.891 to 0.981 for Cu(II)-mixed and others were enhanced to 0.999 for LPSO, the  Table 2 and Figures 5 and 6 clearly verify the effect of linearization. While the fitting accuracy by R 2 increased from 0.891 to 0.981 for Cu(II)-mixed and others were enhanced to 0.999 for LPSO, the accuracy decreased from~0.9 to 0.526 and 0.240 for Cu(II)-single and As(V)-single, respectively, for the LPFO model. Similarly, poor values from linearization were reported in several studies [45,46]. Attempts have been made to overcome the poor relation for the LPFO model. Na and Park (2011) [47] and Simonin (2016) [40] separated the times in terms of the initial from the other times. Both results show that the calculated parameters at the initial time can be relatively compared to the nonlinear equation and PSO; however, the rest of the region should be ignored, indicating that a different adsorption mechanism is expected [47]. In this study, Figure 5 is shown with two regions and Table 2 includes the separate parameters clarified from previous studies. For the condition of 1 ≤ h, the R 2 and q e cal. are comparable to NLFO, but at 3 ≥ h, both R 2 and q e cal. were not comparable Thus, we concluded that linearization of the PFO model included unreasonable data that were influenced by the failure to represent the kinetics and limited by the initial adsorption time [40]. Upon comparison of nonlinear and linear PFO and PSO models, the highest R 2 was acquired for LPSO followed by NPSO and NPFO, and the relative lowest was for the LPFO model. This trend is the same as shown by Kumar (2006), who studied methylene blue on activated carbon [44]. Moreover, the qecal. was compared with qeexp. for all conditions. Based on the results of comparison with R 2 and qe-cal., to estimate the appropriate parameters, the nonlinear application was more suitable and acceptable in this study. Although the comparison of R 2 cannot describe a sufficient criterion [48], Simonin (2016) suggested that the PSO model better describes a diffusion-controlled process than the PFO model [40].
Before discussing the value of K, the effect of the initial concentration should be first considered because a higher initial concentration contributes to a higher concentration gradient, which can enhance the initial sorption rate and require more time to reach equilibrium [49]. However, it can cause a decrease in the entire sorption rate due to the higher competition between adsorbate and adsorbent active sites [50]. In this study, Cu(II) and As(V) in the single and mixed solutions were 0.055 and 0.049 mmol/L and 0.038 and 0.055 mmol/L, respectively, where ~31% of the Cu(II) concentration was decreased in the mixed solution and a ~l0% difference in concentration between Cu(II) and As(V) in single solution was found. According to the K1 and K2 from NPFO, NPSO, and LPSO (LPFO was excluded), the K value was in the following order regardless of kinetic model: As(V)-single > Cu(II)-single > As(V)-mixed >> Cu(II)-mixed. The effect of the initial concentration difference for Cu(II) was negligible because the K value for Cu(II)-mixed showed 3-11 times lower than for other conditions, although the concentration of Cu(II) in the mixed solution showed a 31% decrease.
Based on the order of K, As(V) and single solution had higher rate constants than Cu(II) and the mixed solution. Generally, the value of K increased with increased initial concentration as chemical conditions [49] and shaking speed as physical conditions [50]. The phenomena indicate that for the mixed condition, As(V) and Cu(II) competed with each other at limited activated sites, and the rate Upon comparison of nonlinear and linear PFO and PSO models, the highest R 2 was acquired for LPSO followed by NPSO and NPFO, and the relative lowest was for the LPFO model. This trend is the same as shown by Kumar (2006), who studied methylene blue on activated carbon [44]. Moreover, the q e cal. was compared with q e exp. for all conditions. Based on the results of comparison with R 2 and q e -cal., to estimate the appropriate parameters, the nonlinear application was more suitable and acceptable in this study. Although the comparison of R 2 cannot describe a sufficient criterion [48], Simonin (2016) suggested that the PSO model better describes a diffusion-controlled process than the PFO model [40].
Before discussing the value of K, the effect of the initial concentration should be first considered because a higher initial concentration contributes to a higher concentration gradient, which can enhance the initial sorption rate and require more time to reach equilibrium [49]. However, it can cause a decrease in the entire sorption rate due to the higher competition between adsorbate and adsorbent active sites [50]. In this study, Cu(II) and As(V) in the single and mixed solutions were 0.055 and 0.049 mmol/L and 0.038 and 0.055 mmol/L, respectively, where~31% of the Cu(II) concentration was decreased in the mixed solution and a~l0% difference in concentration between Cu(II) and As(V) in single solution was found. According to the K 1 and K 2 from NPFO, NPSO, and LPSO (LPFO was excluded), the K value was in the following order regardless of kinetic model: As(V)-single > Cu(II)-single > As(V)-mixed >> Cu(II)-mixed. The effect of the initial concentration difference for Cu(II) was negligible because the K value for Cu(II)-mixed showed 3-11 times lower than for other conditions, although the concentration of Cu(II) in the mixed solution showed a 31% decrease.
Based on the order of K, As(V) and single solution had higher rate constants than Cu(II) and the mixed solution. Generally, the value of K increased with increased initial concentration as chemical conditions [49] and shaking speed as physical conditions [50]. The phenomena indicate that for the mixed condition, As(V) and Cu(II) competed with each other at limited activated sites, and the rate declined. As mentioned in Section 3.1, the functional group in chitosan, the amino group, is present in the form of NH 2 or NH 3 + depending on the solution pH, and it has been suggested that Cu(II) can coordinate with NH 2 and As(V) can react with NH 3 + via electrostatic forces [15]. Therefore, electrostatic forces initially occurred and led to the rapid adsorption rate of As(V).

Diffusion Model
Both diffusion and adsorption on the activated site were influenced by the surface area, the reactivity of the surface, and liquid film thickness for external diffusion, and by the reaction of surface and pore structure for internal diffusion. Generally, film diffusion is controlled in a specific system followed at poor mixing, low concentration, small particle sizes of the adsorbent, and a higher affinity of the adsorbate for the adsorbent; whereas, good mixing, larger particle size of the adsorbent, high concentration of adsorbate, and a low affinity of the adsorbate for the adsorbent are controlled at intraparticle diffusion [21,49,51].

Intraparticle Diffusion
A plot of adsorbate uptake versus the square root of time (t 0.5 ) is shown in Figure 7, based on Equation (6), and some calculated parameters are listed in Table 3. There are theoretical interpretations of the intraparticle diffusion equation. C is an arbitrary constant representing the boundary layer thickness, and a larger value of C represents a thicker boundary layer [18,21,52]. If the value of C is zero, which corresponds to no boundary layer, the linear line should pass through the origin. Consequently, film diffusion could be ignored due to no or less thickness, and thus, intraparticle diffusion would remain as the rate-controlling step through the entire adsorption kinetic process. However, this is only the theoretical explanation using Equation (6). Many studies have reported nonzero intercepts, indicating that the rate-limiting step involves both intraparticle and film diffusion in most adsorption processes. As shown in Figure 7 and Table 3, owing to the very low R 2 with the single regression line, it is unable to predict one linear line for the experiment data; instead, it could be divided with two different segments for all cases. Thus, Figure 7 was segregated with two linear regressions, illustrating that both film and intraparticle diffusion control the adsorption diffusion. Moreover, another study [53] suggested that intraparticle diffusion has three regressions in the macro, meso, and micro pores with a horizontal line as the equilibrium. Another study observed three linear regions, including initially rapid surface loading, then pore diffusion, and finally horizontal equilibrium [21]. However, it was not easy to differentiate three or four regions in this study. Consequently, the two linear regions were primarily observed and clearly explained by the adsorption process between the film and intraparticle diffusion [49,54]. Additionally, the initial rapid increase was represented by film diffusion [21,55]. As a result, two linear lines and the value of C (y-axis) both revealed diffusion was coincidentally involved in the adsorption of Cu(II) and As(V).
Some observations were made based on the data in Table 2. The highest and lowest values of K i were for As(V)-single and Cu(II)-mixed in the first regression line, respectively, and the order was reversed to Cu(II)-mixed and As(V)-single in the second line. As mentioned regarding the first rapid adsorption related to film diffusion, the higher K i1 for As(V) compared with that of Cu(II) indicated a higher surface adsorption reaction with an active site leading to an electrostatic force between NH 3 + and HAsO 4 − . According to the assumption where close to the zero of slope (K) in the second section represents intraparticle diffusion, which stands for the equilibrium process [21], the single solutions of Cu(II) and As(V) quickly accomplished equilibrium, whereas mixed solutions required more time to reach equilibrium, which was the same result as with the PSO model. The order of C was As(V), As(V)-Cu(II), Cu(II), and Cu(II)-As(V) regardless of separation, which is related to boundary diffusion or surface adsorption. As a result, the higher C for As(V) demonstrated that As(V) removal was primarily associated with surface adsorption [22].
The order of C was As(V), As(V)-Cu(II), Cu(II), and Cu(II)-As(V) regardless of separation, which is related to boundary diffusion or surface adsorption. As a result, the higher C for As(V) demonstrated that As(V) removal was primarily associated with surface adsorption [22].

External Mass Transfer
To identify the external mass transfer, Equation (7) was expressed with Equation (13) by plotting the values between Ct/Co and ln (Ct/Co), respectively.
Like the intraparticle diffusion model, when the plot using Equation (13) is linear and passes through zero regardless of Ks, A, or V, it suggests that film diffusion governs the adsorption kinetic process [56]. To carefully predict the external diffusion, Figure 8a was segregated with two regions in Figure 8b, and then showed that the slope sharply increased at the beginning (≤ 1 h) and gradually decreased after 1 h, which is a similar trend as that found in a prior study [47]. Note that the simplified Equation (13) is comparable with the LPFO model in Equation (9); additionally, Figure 8 is very similar to Figure 5 in the current study. The rapid slope (Ks•A/V) of the regression line indicates that external mass transfer occurred [57], and the lowest slope for the Cu(II) mixed solution can be assumed to be relatively associated with a strong effect of intraparticle diffusion. Therefore, both intraparticle and external diffusion simultaneously influence the adsorption kinetics.

External Mass Transfer
To identify the external mass transfer, Equation (7) was expressed with Equation (13) by plotting the values between C t /C o and ln (C t /C o ), respectively.
Like the intraparticle diffusion model, when the plot using Equation (13) is linear and passes through zero regardless of K s , A, or V, it suggests that film diffusion governs the adsorption kinetic process [56]. To carefully predict the external diffusion, Figure 8a was segregated with two regions in Figure 8b, and then showed that the slope sharply increased at the beginning (≤ 1 h) and gradually decreased after 1 h, which is a similar trend as that found in a prior study [47]. Note that the simplified Equation (13) is comparable with the LPFO model in Equation (9); additionally, Figure 8 is very similar to Figure 5 in the current study. The rapid slope (K s ·A/V) of the regression line indicates that external mass transfer occurred [57], and the lowest slope for the Cu(II) mixed solution can be assumed to be relatively associated with a strong effect of intraparticle diffusion. Therefore, both intraparticle and external diffusion simultaneously influence the adsorption kinetics.

Conclusions
In this study, we described the interaction between the amino group in the chitosan polymer and the cation of Cu(II) and anion of As(V) to determine the adsorption mechanism. Additionally, the effect of each ion on adsorption was also studied for single and mixed solutions. According to the increased capacity of the Cu(II) and As(V) mixed solution, the amino group in chitosan was in the form of NH2 and NH3 + , and each functional group directly participated in the removal of the anion and cation, respectively. The PSO model is more suitable than PFO, indicating that surface chemisorption occurred primarily for both Cu(II) and As(V) adsorption. Because the linearization of PFO expanded the error, lowering the correlation coefficient (R 2 ), two segregation regression lines dependent on time were effective at describing the adsorption. Because the two types of diffusion models showed no passing through of the zero regression line, both intraparticle and film diffusion were considered as rate-controlling steps for Cu(II) and As(V) adsorption. A higher rate constant (K) for As(V) than for Cu(II) was obtained from the electrostatic interaction between NH3 + and HAsO4 2− .

Conclusions
In this study, we described the interaction between the amino group in the chitosan polymer and the cation of Cu(II) and anion of As(V) to determine the adsorption mechanism. Additionally, the effect of each ion on adsorption was also studied for single and mixed solutions. According to the increased capacity of the Cu(II) and As(V) mixed solution, the amino group in chitosan was in the form of NH 2 and NH 3 + , and each functional group directly participated in the removal of the anion and cation, respectively. The PSO model is more suitable than PFO, indicating that surface chemisorption occurred primarily for both Cu(II) and As(V) adsorption. Because the linearization of PFO expanded the error, lowering the correlation coefficient (R 2 ), two segregation regression lines dependent on time were effective at describing the adsorption. Because the two types of diffusion models showed no passing through of the zero regression line, both intraparticle and film diffusion were considered as rate-controlling steps for Cu(II) and As(V) adsorption. A higher rate constant (K) for As(V) than for Cu(II) was obtained from the electrostatic interaction between NH 3 + and HAsO 4 2− .
Funding: This research was supported by the National Research Foundation of Korea (NRF-2019R1A2C1009129).

Conflicts of Interest:
The author declares no conflict of interest.

Nomenclature
A Surface area of adsorbent (m 2 ) C 0 Initial concentration in the solution, mg/L C t Liquid phase concentration at any time, mg/L K 1 Sorption rate constant (min −1 ) at PFO K 2 Sorption rate constant (g/mg·min) at PSO K i Intraparticle diffusion rate constant (mg/g· min 1/2 ) K n n order diffusion rate constant K s The liquid-film mass transfer coefficient (m 2 /s) M Mass of adsorbent (g) n Kinetic order q e Adsorbate uptake at equilibrium (mg/g) q t Adsorbate uptake at any time (mg/g) t Reaction time (min) V Volume of batch experiment (L)