Synthesis of chitosan zero-valent iron nanoparticles-supported for cadmium removal: characterization, optimization and modeling approach

Heavy metals are highly toxic and hazardous elements that have a high atomic weight and a density at least 5 times greater than that of water. They are widely used in industrial, domestic, agricultural, medical and technological applications, which has led to their continuous release into the environment. Due to their high degree of toxicity, arsenic (As), cadmium (Cd), chromium (Cr), lead (Pb) and mercury (Hg) rank among the priority metals that are of public health significance (Jaafarzadeh et al. 2012; Begum et al. 2016). Cadmium (Cd²⁺) is one of the most dangerous pollutants that is released into the environment, mainly via industrial applications such as phosphate fertilizers, batteries, electroplating industries, mining, metal production, stabilizers and alloys and the manufacturing of pigments. It has been classified as a human carcinogen and teratogen impacting lungs, kidneys, liver and reproductive organs (Azari et al. 2015; Naghizadeh 2015). The World Health Organization (WHO) has set a maximum guideline concentration of 0.003 mg/L for Cd²⁺ in drinking water (WHO 2007). Considering the negative effects, toxicity and stability of heavy metals, their complete removal from water resources and wastewater effluents is deemed necessary.

In the last several years, different technologies have been studied to remove heavy metals from aqueous solutions including adsorption, ion exchange, chemical precipitation, membrane filtration and coagulation–flocculation (Azari et al. 2015). However, most of them suffer from several disadvantages such as higher operational and capital costs, more energy and chemicals consumption, and problems regarding sludge disposal (Sobhanardakani et al. 2015). Other drawbacks are the requirement for large settling tanks in chemical precipitation, regeneration in the ion exchange process, chemical requirements, low efficiency in coagulation–flocculation methods and large amounts of sludge in membrane filtration (Gupta et al. 2011b; Kakavandi et al. 2015). During the past few years, the adsorption process has been widely applied; also, this process is proven to be a suitable method for the treatment of heavy metals (Ahmadi et al. 2011; Amiri et al. 2011). In this regard, up to now, a wide variety of adsorbents have been used for Cd²⁺ removal such as agricultural waste biomass, chitosan–silica, microorganisms, biopolymers, zeolites, metal oxides, fly ash and activated carbon (Jaafarzadeh et al. 2014; Lim & Aris 2014).

However, most of these adsorbents showed a relatively low adsorption capacity for Cd²⁺ under the optimum operation conditions. In addition, some operational problems such as resultant turbidity in the treated water or effluent, and consequently the need to filter or centrifuge, have limited the application of these adsorbents, particularly nano-sized adsorbents. Magnetic nanoparticles (e.g. NZVI, Fe₃O₄, α-Fe₂O₃, γ-Fe₂O₃ and FeO(OH)) have recently been adopted by researchers in the field of adsorption/biosorption for removing pollutants from aquatic environments, which makes separation of both adsorbent and adsorbate much easier (Mohseni-Bandpi et al. 2015). Several authors have magnetized adsorbents such as activated carbon for Pb²⁺ and Hg adsorption (Oliveira et al. 2002; Kakavandi et al. 2015), carbon nanotubes for Pb²⁺, Ni and Sr adsorption (Chen et al. 2009; Hu et al. 2010), zeolite for Cr, Cu, and Zn adsorption (Oliveira et al. 2004) and CS for Zn²⁺ and Pb²⁺ adsorption (Fan et al. 2011, 2013) by magnetic iron nanoparticles as a magnetic separation technology.

Among magnetic nanoparticles, NZVI has been applied recently for in-situ and ex-situ remediation, due to being non-toxic and inexpensive (Esfahani et al. 2014a). NZVI, due to its extremely small particle size, large specific surface area and greater reactive sites and capacity, is notable for this purpose in wastewater treatment to remove heavy metals with a higher efficiency (Esfahani et al. 2014b). Moreover, the magnetic properties of NZVI facilitate the rapid separation of nano iron from soil and water via a magnetic field (Babaei et al. 2015a). However, there is a strong tendency of NZVI particles to agglomerate as well as to become oxidized, resulting in a reduction in surface area, reactivity and removal efficiency (Babaei et al. 2015a). An effective approach to overcome this problem is to incorporate NZVI into a porous supporting material. Recent studies have reported that NZVI particles can be coated with CS (a protective polymer due to its outstanding chelation behavior) to increase its dispersibility and stability (Liu et al. 2012). Furthermore, these supports can facilitate the separation of NZVI particles from aqueous solutions.

Herein, we hypothesize that NZVI particle impregnation on the CS surface combines the synergistic effects of NZVI and CS, which may have a superbly enhanced adsorption activity as well as easy separation. The present study therefore aimed to synthesize CS@NZVI using a liquid phase method. The influence of operating parameters in the adsorptive removal of Cd²⁺ was evaluated in details in a batch system. Isothermic and kinetic studies were also carried out under optimum conditions. Finally, the regeneration and reusability of the composite were indeed evaluated for three consecutive cycles.

All chemicals were of analytical laboratory grade and used without further purification. Sodium borohydride (NaBH₄) was purchased from Sigma-Aldrich. Cadmium nitrate tetrahydrate (Cd(NO₃)₂.4H₂O, Merck, Co) was used for preparing the stock solutions of Cd²⁺ according to the ASTM D3557-12 (ASTM 2012) procedure. The pH of the solutions was adjusted by adding 0.1 M hydrochloric acid (HCl) and sodium hydroxide (NaOH) solutions. All the reagents were prepared with de-ionized water (DI-water) and kept in a refrigerator at 4 °C prior to experiments.

CS@NZVI composite was synthesized in the laboratory using a chemical reduction method (reducing Fe³⁺ to Fe⁰ using NaBH₄). Excess NaBH₄ was used to ensure that all the Fe³⁺ was reduced. Firstly, 0.25 g of CS was dissolved in 50 mL of 0.05 M acetic acid. Due to the poor solubility of CS, the mixture was vortexed to aid complete dissolution and kept for 2 h at 150 rpm. To this solution, 1 g of FeCl₃.7H₂O was added and the solution was stirred quickly in an N₂-purged environment for 2 h. Then, to this mixture freshly prepared aqueous solution containing 2% NaBH₄ was added drop-wise. At this stage, black precipitation was observed, and evolution of H₂. Again, the mixture was stirred for another 60 min until the entire reduction of metal salts. The black solid was collected using a magnet (with a 1.5 tesla filed magnet) and washed at least three times with oxygen-free DI-water to get rid of the extra chemicals. The CS@NZVI composite was dried at 100 °C for 4 h, and stored in a brown sealed bottle under dry conditions for characterization and future use (Geng et al. 2009; Gupta et al. 2011a).

Batch experiments for the adsorption of Cd²⁺ on CS@NZVI composite were carried out in 250 mL polytetrafluoroethylene bottles filled with 50 mL of the pH-adjusted Cd²⁺ solutions at 25 ± 1 °C. The effects of experimental parameters such as the pH of the solution, contact time, different CS@NZVI and Cd²⁺ concentrations and solution temperatures on the removal efficiency of Cd²⁺ were investigated. After adjusting the pH of the solution, a specific amount of composite was put in the aqueous solution, having a fixed concentration. Then, bottles were agitated on a rotary shaker at a rate of 200 rpm and maintained for a certain period of time at a constant temperature (25 ± 1 °C). At appropriate time intervals, 2 mL of the solution was withdrawn from each bottle and the composite was magnetically separated using a strong magnet. After that, the remaining Cd²⁺ concentration in the solution was determined according to the ASTM (D3557-90 method) (ASTM 2012) using atomic absorption spectrophotometry (Analytikjena, vario 6, Germany) at a wavelength of 228.8 nm. Herein, all measurements were performed in an air/acetylene flame. The lamp current and slit width were 2.0 mA and 1.2 nm, respectively. The instrument was calibrated with a standard solution (in the range 0.05–2.0 mg/L) within a linear range, and a high correlation coefficient (R² > 0.997) was obtained. All experiments were performed in duplicate and the results were reported as the mean values of measurements.

The results of TEM analysis, Figure 1(b), showed that NZVI had a diameter less than 50 nm and also demonstrated that it was successfully synthesized as individual nano-sized particles. The specific surface area, volume, and average pore diameter of CS@NZVI were measured using the BET method. The surface of the synthesized adsorbent, according to this analysis, was 78.3 m²/g. It is notable that the specific surface area of CS decreased after the coating of NZVI, as reported in the literature (Babaei et al. 2015a). This decrease may result from the impregnation process and/or NZVI presence in the structure of CS. Similar observations were also reported by other researchers (Kakavandi et al. 2014, 2015). The average size and volume pores of the composite were obtained to be 26.57 nm and 0.982 cc/g, respectively. According to the IUPAC classification, the average size of 26.57 nm can be classified as mesoporous groups (Depci 2012). The results of this analysis reveal that the CS@NZVI is porous in structure and could provide more reactive sites and a good adsorption capacity for contaminants. Because adsorption reactions mainly occurred on the adsorbent surfaces, the functional groups on the surfaces of the adsorbent can play a significant role in the adsorption process. To characterize the functional groups on the surfaces of the adsorbent and to measure the binding mechanism of the pollutants, the FTIR spectra of the CS@NZVI before and after adsorption of Cd²⁺ in the range of 400–4,000 cm⁻¹ are shown in Figure 1(c) and 1(d), respectively.

The FTIR spectra showed some absorption peaks belonging to various functional groups or different vibration modes. A comparison between the FTIR spectrums of the CS@NZVI before and after the adsorption of Cd²⁺ is given in Table 1. The absorption bonds at wave number (ν) values at ∼3,354 and ∼3,268 cm⁻¹ indicate the presence of O–H and N–H bond stretching, respectively. The absorption peaks at 2,860 cm⁻¹ are due to the C–H stretching vibration of the –CH₂ groups in CS (Du et al. 2014; Mohseni-Bandpi et al. 2015). The peak observed at 1,631 cm⁻¹ may be from the N-H bending vibration, indicating the existence of amide(II) and hydroxyl groups in CS (Liu et al. 2012). Moreover, the bond at near 1,600 cm⁻¹ that appeared on CS@NZVI before and after adsorption of 10 mg/L Cd²⁺ was assigned to the OH bending vibrational mode due to the adsorption of moisture when FTIR sample disks were prepared in an open-air atmosphere (Mohseni-Bandpi et al. 2015). The bands at about 1,363 cm⁻¹ can be attributed to C–N stretching vibration (Malkoc & Nuhoglu 2006). In the FTIR spectra, the peaks at around 1,140 cm⁻¹ can be apportioned to the C= O stretching of ether groups (Malkoc & Nuhoglu 2006). The peaks at 1,083 cm⁻¹ and 1,023 cm⁻¹ correspond to C–OH bond stretching (Reddy & Lee 2013). The peaks at around 570 cm⁻¹ in the CS@NZVI spectrum were attributed to the Fe–O stretching vibration, implying that the NZVI nanoparticles were successfully prepared and introduced into the CS (Yang et al. 2014).

In the CS@NZVI spectrum after adsorption, a significant reduction of absorption in this spectral area can be attributed to the formation of CS − Fe bonds. All the aforementioned peaks were also observed in the ‘after adsorption’ FTIR spectra with notable changes. These functional groups may form surface complexes with Cd²⁺ and thus can increase the specific adsorption of Cd²⁺ by CS@NZVI. As shown in Figure 1(c) and 1(d) and Table 1, the spectra display a number of absorption peaks, indicating the complex nature of the CS@NZVI. Large changes are clearly observed on the FTIR spectrum of CS@NZVI following Cd²⁺ adsorption. After Cd²⁺ adsorption, the FTIR spectrum, Figure 1(d), shows a new strong peak at 2,868 cm⁻¹, belonging to the stretching vibration of symmetric and asymmetric –CH₂ groups (Ngah et al. 2008). Furthermore, FTIR spectra of Cd²⁺ adsorbed on CS@NZVI indicated that the peaks expected at 3,354, 3,268, 2,856, 1,589, 1,363 and 1,147 cm⁻¹ had shifted, respectively to 3,357, 3,288, 2,868, 1,597, 1,376 and 1,151 cm⁻¹ due to Cd²⁺ sorption. It seems that the mentioned functional groups influence the Cd²⁺ adsorption on the CS@NZVI. Generally, the findings of FTIR studies clearly confirm the existence of CS and NZVI in the CS@NZVI composite.

Liu et al. (2013) studied the Cd²⁺ adsorption on CS beads-supported Fe⁰ and showed that when solution pH increased, the number of negatively charged sites was improved, leading to the enhanced attraction force between heavy metals (Cu²⁺, Cd²⁺ and Pb²⁺) and the beads surface. Furthermore, Azari et al. (2015) reported that as the pH increased, surface positive charges of the adsorbent decreased and the more active surface sites can be obtained for Cd²⁺, which resulted in lower repulsion of the adsorbing metal ions. At alkaline conditions, however, a decrease in the adsorption efficiency can be derived from the formation of metal hydroxides precipitation and also a decrease in the concentration of Cd²⁺, as reported in the literature (Kakavandi et al. 2015). Rao et al. (2009) reported that at pH > 7.5, the predominant species of Cd exists in the hydrolyzed form (i.e. Cd(OH)⁺ and Cd(OH)₂⁰) and Cd²⁺ ions are present in only very small amounts. Therefore, at the value of pH < 7.0, the main species adsorbed onto the CS@NZVI were predominantly Cd²⁺ and less amounts of Cd(OH)⁺ and Cd(OH)₂. Based on the aforementioned, at the optimum pH the predominant species of Cd were in ionic form (Cd²⁺) and metal hydroxide precipitation does not take place.

However, a decrease in adsorption capacity with an increase in the adsorbent dosage is mainly attributed to the increase in unsaturation of adsorption sites through the adsorption reaction (Jafari et al. 2016; Kakavandi et al. 2016a). In addition, some of the particle interactions (e.g. aggregation) which result from a high sorbent concentration lead to a significant reduction in the active surface area of the adsorbent and, consequently, reduce its adsorption capacity. Similar observations have been reported for adsorption of Cd²⁺ onto the different adsorbents in the literature (Rao et al. 2009; Shen et al. 2009; Azari et al. 2015).

The effect of initial concentrations of Cd²⁺ on its removal efficiency by CS@NZVI in the range of 10-300 mg/L is shown in Figure 3(b). It can be seen that the removal efficiency decreased with enhancement of the Cd²⁺ from 10 to 300 mg/L. So that, with the rise in the initial concentration from 10 to 300 mg/L, the removal efficiency decreased from 99.8% to 34.4%. This is probably due to the fixed number of active sites on the adsorbent versus the number of metal ion molecules (Teymouri et al. 2013). Figure 3(b) also reveals that the adsorption capacity of Cd²⁺ on the CS@NZVI significantly enhanced as the initial Cd²⁺ concentration increased. This phenomenon can be described by the fact that amounts of Cd²⁺ adsorbed per unit mass of CS@NZVI increase with an increase in initial Cd²⁺ concentration in the solution. Moreover, an increase in initial concentration dramatically enhanced the interaction between the adsorbent and Cd²⁺. This can be attributed to the increased force of concentration gradient (Kumar et al. 2009).

Generally, at higher initial concentration of metal ions the available adsorption sites of the CS@NZVI become fewer and the percent removal of metal ions is dependent upon the initial concentration. However, the ratio of the initial number of metal ions to the available sorption sites of the CS@NZVI is decreased at a lower initial concentration of Cd²⁺, and subsequently the fractional adsorption of metal ions by the CS@NZVI becomes independent of its initial concentration (Rao et al. 2009; Yang et al. 2014; Azari et al. 2015).

In Table 2, the values of the kinetic model parameters of Cd²⁺ adsorption onto CS@NZVI are listed. In this study, we used four widely used kinetic models: pseudo-first-order, pseudo-second-order, Elovich, and intraparticle diffusion models to estimate overall sorption rates. Further details of these models (i.e. equations and parameters) are given in the supplementary data, Table S1 (available with the online version of this paper). The correlation coefficients were found to be less than 0.96, 0.85 and 0.67 for the pseudo-first-order, Elovich and intraparticle diffusion kinetic models, respectively; whereas the corresponding amount calculated for the pseudo-second-order kinetic model was more than 0.999. This suggests that the pseudo-second-order is a better fit to the experimental data of Cd²⁺ adsorption with a significantly high coefficient of correlation (R²) (>0.99), compared to other kinetic models. For the pseudo-second-order model it is also strongly confirmed that the calculated q_e values are in good agreement with the experimental q_e values, indicating that this model better explains the adsorption process of Cd²⁺ on the CS@NZVI than the other models. The confirmation of this model demonstrates that the concentrations of both adsorbent and adsorbate are associated with the rate-determining step of the adsorption process (Jafari et al. 2016). It also suggests that chemisorption was the rate-limiting step in the adsorption process of Cd²⁺ onto the CS@NZVI, and there was no mass transfer reaction (Kakavandi et al. 2014). In the previous studies conducted, the same model for the adsorption of Cd²⁺ on various adsorbents, such as magnetic activated carbon (Azari et al. 2015), activated carbon (Rao et al. 2009) (Dong et al. 2014), and clarified sludge (Naiya et al. 2008) were reported.

Other models (i.e. pseudo-first-order, Elovich and intraparticle diffusion) present lower R² values, indicating that these models could not properly fit the experimental kinetic data. Based on the results, it was found that the intraparticle diffusion model plays a less significant role in the adsorption process. According to Table 2, for intraparticle diffusion the y-intercept (C_i) is not zero, illustrating that the intraparticle diffusion is part of the adsorption but not the only rate-controlling step in this process, as reported previously by Boparai et al. (2011). Therefore, it can be stated that other mechanisms (i.e. complexes or ion-exchange) could also control the rate of the adsorption of Cd²⁺ on CS@NZVI.

In this study, Langmuir, Freundlich, Temkin and Dubinin–Radushkevich (D-R) equilibrium as the four most common isotherm models were employed to predict the behavior of Cd²⁺ adsorption onto the CS@NZVI surfaces. The equations and corresponding parameters of the aforementioned models are represented in Table S1. The adsorption isotherm experiments were conducted using 10 to 300 mg/L Cd²⁺ under the optimum conditions (i.e. pH 7.0 ± 0.2, 0.6 g/L adsorbent and 90 min contact time) at 25 ± 1 °C. Table 1 shows the obtained values of equilibrium isotherm parameters of the Cd²⁺ adsorption onto the CS@NZVI surfaces. Based on the correlation coefficients (R²), the adsorption isotherm models fitted the experimental data in accordance to the following order: Freundlich > D-R > Langmuir > Temkin. Considering this result, the Freundlich model is a better fit to the experimental data of the Cd²⁺ adsorption by CS@NZVI than the other three models. In addition, we observed the best fit for the Freundlich model by employing a nonlinear method, as plotted in Figure 4(b). This model suggests that the heterogeneous functional sites are distributed uniformly on the surfaces of CS@NZVI and the adsorption of Cd²⁺ ions onto non-energetically equivalent sites of the CS@NZVI (Kakavandi et al. 2015; Rezaei Kalantry et al. 2016). Meanwhile, the value of 1/n (less than unity) in the Freundlich isotherm model implies the favorable adsorption of Cd²⁺ onto CS@NZVI. In addition, as presented in Table 2, the values for the dimensionless separation parameter R_L (R_L = 1/(1 + k_LC₀)), which were related to the Langmuir model, fell between 0 and 1. Since R_L> 1, R_L = 1, R_L = 0 and 0 < R_L< 1 indicate unfavorable, linear, irreversible and favorable adsorption, respectively, it can be concluded that the simultaneous adsorption of Cd²⁺ onto CS@NZVI is favorable.

For the D-R model, the mean free energy of adsorption (E = 1/(-2D)^0.5) per mole of the adsorbate is the energy needed to transfer one mole of an adsorbate to the adsorbent surfaces from infinity in solution. It gives information about either chemical or physical adsorption. With the magnitude of E, between 8 and 16 kJ/mol, the adsorption mechanism follows chemical ion-exchange, while for the values of E < 8 kJ/mol, the adsorption process is of a physical nature (Azouaou et al. 2010; Kakavandi et al. 2015). As shown in Table 2, the value of the mean free energy of adsorption, E, for Cd²⁺ on CS@NZVI, was found to be between 8 and 16 kJ/mol, indicating that the adsorption process follows a chemical mechanism. The chemisorption nature of Cd²⁺ adsorption on different types of adsorbents has been reported previously (Ünlü & Ersoz 2007; Naiya et al. 2008; Boparai et al. 2011).

The maximum adsorption capacity, q_m, of the CS@NZVI was compared with the other adsorbents (see Table 3). It is worth mentioning that the CS@NZVI poses a better adsorption capacity, compared with the capacity of other adsorbents applied in previous research. The observed differences in the adsorption capacities for the listed adsorbents can be due to the structure, surface area and the properties of the functional groups in each adsorbent. As presented in Table 3, α-ketoglutaric acid-modified magnetic CS provides a high adsorption capacity for Cd²⁺ compared with other adsorbents, which can be attributed to its textural characteristics, high porosity and surface area and functional groups. Moreover, it is worth noting from this table that the CS@NZVI had a positive effect on Cd²⁺ removal and can be considered as one of the most effective adsorbents for Cd²⁺ adsorption. Nevertheless, in order to enhance the adsorption capacity of CS@NZVI, further studies can be conducted on its modification through increasing the surface area and changing of functional groups.

Results revealed that CS@NZVI has a high potential and adsorption capacity for Cd²⁺ ion removal from aqueous solutions. At a pH of 7 ± 0.2, the adsorption efficiency was enhanced by an increase in the contact time and adsorbent dosage and a decrease in the initial Cd²⁺ concentration. The equilibrium adsorption data were found to fit best using a Freundlich isotherm and pseudo-second-order kinetic models. The maximum adsorption capacity obtained was 142.8 mg/g based on the Langmuir isotherm. The adsorption process of Cd²⁺ onto the synthesized composite was chemisorption. Moreover, the adsorbent was successfully recycled for three cycles with a little decrease of variation in adsorption ability. The CS@NZVI provides very promising results for cost-effective treatment of wastewaters contaminated by Cd²⁺, as well as high adsorption capacities, good and rapid separations and an efficient technology for heavy metals removal.

The present work was financially supported by the Environmental Technologies Research Center, Ahvaz Jundishapur University of Medical Sciences (Grant No. ETRC-9112). The authors are grateful for the support of Iranian Nano Technology Initiative Council.