An Investigation into the Adsorption of Ammonium by Zeolite-Magnetite Composites

: The discharging of ammonium from industrial, domestic, and livestock sewage has caused eutrophication of the water environment. The objectives of this study are to synthesize magnetic zeolite (M-Zeo) by an eco-friendly, economical, and easy procedure and to investigate its suitability as an adsorbent to remove ammonium from an aqueous solution. Based on characterization from XRD, BET, and SEM-EDS, Fe 3 O 4 was proved to successfully load on natural zeolite. The effect of pH, temperatures, reacting times, initial ammonium concentrations, and regeneration cycles on ammonium adsorption was examined by batch experiments. The ammonium adsorption process can be best described by the Freundlich isotherm and the maximum adsorptive capacity of 172.41 mg/g was obtained. Kinetic analysis demonstrated that the pseudo-second-order kinetic model gave the best description on the adsorption. The value of pH is a key factor and the maximum adsorption capacity was obtained at pH 8. By using a rapid sodium chloride regeneration method, the regeneration ratio was up to 97.03% after ﬁve regeneration cycles, suggesting that M-Zeo can be recycled and magnetically recovered. Thus, the economic-efﬁcient, great ammonium afﬁnity, and excellent regeneration characteristics of M-Zeo had an extensively promising utilization on ammonium treatment from liquid.


Introduction
Nitrogen compounds, ammonium or ammonia generated from population growth, agriculture, and food predicting, in particular, in sufficient concentration can promote water eutrophication. The presence of ammonium in industrial, domestic, and livestock sewage has always been a major concern. The increasing amount of ammonium in wastewater requires efficient sewage treatment technologies, including air stripping, biological treatment, electrochemical treatment, membrane distillation, struvite precipitation, microwave radiation, absorption, and ion exchange [1][2][3][4]. Biological treatment is considered the most efficient method on ammonium removal from wastewater with high nitrogen concentration. However, the traditional steps for ammonium removal in biological treatment comprise aerobic nitrification and anoxic denitrification, which consumes abundant energy [5,6]. In addition, large structure floor area, sensitive to shock load, high capital investment and operation cost, and complex management limit its large-scale practical application. More importantly, the current sewage treatment system cannot eliminate the strong disturbance of human activities to the nitrogen compounds in the natural process, and still lead to many nitrogen compounds finally discharged into the water environment. Actually, ammonium or ammonia is also a resource and should be recovered from wastewater. Dawson by using ethanol and deionized water for a few times until the conductivity was below 10 µs/cm. The M-Zeo was obtained after drying at 40 • C for 24 h.
Except zeolite, other reagents (such as ferric chloride, ammonium chloride, ethylene glycol, and so on) employed in the current contribution were all of an analytical reagent and purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China).

Batch Adsorption Experiments
The influence of pH on the ammonium adsorptive capacity by M-Zeo was tested under ambient conditions. NH 4 + -N solutions (50 mL, 1000 mg/L) were put into 100 mL polyethylene centrifuge tubes with caps, then, the pH value was regulated to 4, 5, 6, 7, 8, and 9 by using 1M HCl or 1M NaOH solutions. A total of 0.02 g adsorbent (M-Zeo) was respectively added into NH 4 + -N solutions and then mingled in thermostatic shakers for 12 h (200 rpm, 25 • C). After equilibrium, the solid phase was separated from the liquid phase by centrifuging, filtering (membrane filter, 0.45 µm), or magnetic attraction, and then NH 4 + -N concentrations were tested. NH 4 + -N adsorption isotherms were performed in thermostatic shakers for 12 h at the desired temperatures (298 K, 308 K, and 318 K). A total of 0.02 g adsorbent (M-Zeo) was respectively added into the NH 4 + -N solutions (50 mL) at pH = 8 and initial concentrations at a range from 5 to 1000 mg/L (5, 50, 100, 200, 500, and 1000 mg/L).
The impact of the regeneration cycles on the adsorption capacity of ammonium was performed at pH=8 and at 298 K. The adsorbents after adsorption were collected and regenerated by using NaCl solutions (50 mL, 2 mol/L). After washing with deionized water for a few sequences, the regenerated M-Zeo was dried at 60 • C for 24 h and then reused to adsorb ammonium from the aqueous solution.

Analysis Methods
The NH 4 + -N concentrations were analyzed by the Nessler's reagent spectrophotometry method with a spectrophotometer (722E, Spectrum Co., Shanghai, China). The mineral phases were performed by X-ray diffraction (XRD) analysis using an X-Ray diffractionmeter (SmartLab, Rigaku, Japan) with a Cu-target and a range of 5-70 • at a scan rate of 10 • min −1 . The morphology and nanostructures of M-Zeo were analyzed by a field emission scanning electron microscope (FESEM, Sigma 300, Zeiss Ltd., Cambridge, UK) with an electron acceleration voltage of 10 kV. The analysis of surface area and pore circumstances of M-Zeo were performed by using a surface area and pore size analyzer (Quanta NOVA 3000e, Quantachrome, Shanghai, China). Figure 1 illustrates the XRD patterns of N-Zeo and M-Zeo. The characteristic diffraction peaks of clinoptilolite appeared on 2θ = 9.76, 11.16, 22.36, 28.04, 30.02, and 31.92 • . The reflection at 2θ = 26.56 • was found and authenticated as quartz according to the standard database. Clinoptilolite was the main phase coexisting with quartz in adsorbents, but the intensities of quartz and clinoptilolite superficially became weaker after modification. As a matter of fact, the diffraction patterns at 2θ = 35.42, 57.28, and 62.48 • identified as magnetite show strong diffraction peaks in the sample of M-Zeo, which proved magnetite as a new phase successfully loaded on zeolite.

Characterization
The SEM images of adsorbents used in the present study are presented in Figure 2. Plate-like morphology crystals, flat surfaces, and massive pores of the zeolite and the channels inside of the zeolite framework can be observed in Figure 2a Figure 2b. The EDS analysis (Figure 2c) also proved that the surface of zeolite was coated with plenty of magnetite. Figure 1 illustrates the XRD patterns of N-Zeo and M-Zeo. The characteristic diffraction peaks of clinoptilolite appeared on 2θ = 9.76, 11.16, 22.36, 28.04, 30.02, and 31.92°. The reflection at 2θ = 26.56° was found and authenticated as quartz according to the standard database. Clinoptilolite was the main phase coexisting with quartz in adsorbents, but the intensities of quartz and clinoptilolite superficially became weaker after modification. As a matter of fact, the diffraction patterns at 2θ = 35.42, 57.28, and 62.48° identified as magnetite show strong diffraction peaks in the sample of M-Zeo, which proved magnetite as a new phase successfully loaded on zeolite. The SEM images of adsorbents used in the present study are presented in Figure 2. Plate-like morphology crystals, flat surfaces, and massive pores of the zeolite and the channels inside of the zeolite framework can be observed in Figure 2a. Many Fe3O4 particles with a scale of 200-300 nm were loaded on the surface of zeolite after magnetic modification in Figure 2b. The EDS analysis (Figure 2c) also proved that the surface of zeolite was coated with plenty of magnetite.  Figure 1 illustrates the XRD patterns of N-Zeo and M-Zeo. The characteristic diffraction peaks of clinoptilolite appeared on 2θ = 9.76, 11.16, 22.36, 28.04, 30.02, and 31.92°. The reflection at 2θ = 26.56° was found and authenticated as quartz according to the standard database. Clinoptilolite was the main phase coexisting with quartz in adsorbents, but the intensities of quartz and clinoptilolite superficially became weaker after modification. As a matter of fact, the diffraction patterns at 2θ = 35.42, 57.28, and 62.48° identified as magnetite show strong diffraction peaks in the sample of M-Zeo, which proved magnetite as a new phase successfully loaded on zeolite.

Characterization
(a) The BET-specific surface area, pore volume, and pore size are templated in Table 1. M-Zeo had better BET-specific surface area and pore volume than N-Zeo. The increase in BET of M-Zeo was attributed to the nanoscale of magnetite occupied on the surface of zeolite. The adsorption curves of both N-Zeo and M-Zeo showed a shape of IV isotherm The BET-specific surface area, pore volume, and pore size are templated in Table 1. M-Zeo had better BET-specific surface area and pore volume than N-Zeo. The increase in BET of M-Zeo was attributed to the nanoscale of magnetite occupied on the surface of zeolite. The adsorption curves of both N-Zeo and M-Zeo showed a shape of IV isotherm (Figure 3), implying the characteristic feature of mesoporous materials. Thus, these mesoporous materials provide more internal specific surface and pore volume [34].  The BET-specific surface area, pore volume, and pore size are templated in Table 1. M-Zeo had better BET-specific surface area and pore volume than N-Zeo. The increase in BET of M-Zeo was attributed to the nanoscale of magnetite occupied on the surface of zeolite. The adsorption curves of both N-Zeo and M-Zeo showed a shape of IV isotherm (Figure 3), implying the characteristic feature of mesoporous materials. Thus, these mesoporous materials provide more internal specific surface and pore volume [34].

Influence of pH on Ammonium Removal
The impact of pH values on NH 4 + -N adsorption by using M-Zeo as an adsorbent was performed by adjusting pH values in a range from 4.0 to 9.0. As shown in Figure 4, among the initial pH values, the maximum ammonium adsorption takes place at pH 8.0. The results proved that the adsorption process in the current study was pH-dependent. The adsorption capacity of ammonium clearly rose from 73.84 mg/g to 140.97 mg/g in the pH range from 4.0-8.0, and then declined to 115.94 mg/g at pH 9.0. The present study observed the same trend in a pH rise with other literatures [11,27,35]. The balance between NH 3 and NH 4 + is pH and temperature-dependent, and the relationship between ammonium and ammonia in an aqueous solution can be expressed as follows:  [36]. pKa can be stated in a temperature relation formula ( • C), represented in Equation (2). Therefore, pKa was calculated as 9.24 according to the following Equation (2). This was in accordance with ammonium existing as NH 4 + at pH 2-8 and NH 3 at pH 10-13 in the aqueous solution (Figure 4b). At a lower pH, zeolite is highly selective for H + and NH 4 + and could be favorable for adsorbing NH 4 + on the external surface of the zeolite. At a higher pH, the decrease of ammonium adsorption capacity can be attributed to the alkaline condition promoting NH 4 + turning into NH 3 .

Adsorption Isotherms and Thermodynamics
Three adsorption models of the Langmuir model, the Freundlich model, and the D-R isotherm were used to fit the process of the M-Zeo adsorbing ammonium [37][38][39]. The Langmuir model is described as Equation (3): here, qm is the maximum adsorptive capacity (mg/g); k (L/mg) refers to the Langmuir constant; Ce is the equilibrium concentration (mg/L); qe is the adsorption capacity on adsorbent (mg/g). The Freundlich model is represented as Equation (4) here, Kf (mg/g) refers to the Freundlich constant. 1/n is a heterogeneous factor, which is involved in the adsorption intensity or surface heterogeneity.
The D-R isotherm is given as Equation (5) here, β is the adsorption energy constant (mol 2 /J 2 ); ε is Polanyi potential, which can be calculated as Equation (6): (b) distribution of ammonium species in liquid.

Adsorption Isotherms and Thermodynamics
Three adsorption models of the Langmuir model, the Freundlich model, and the D-R isotherm were used to fit the process of the M-Zeo adsorbing ammonium [37][38][39]. The Langmuir model is described as Equation (3): here, q m is the maximum adsorptive capacity (mg/g); k (L/mg) refers to the Langmuir constant; C e is the equilibrium concentration (mg/L); q e is the adsorption capacity on adsorbent (mg/g). The Freundlich model is represented as Equation (4): here, K f (mg/g) refers to the Freundlich constant. 1/n is a heterogeneous factor, which is involved in the adsorption intensity or surface heterogeneity. The D-R isotherm is given as Equation (5): ln q e = ln q m − βε 2 (5) here, β is the adsorption energy constant (mol 2 /J 2 ); ε is Polanyi potential, which can be calculated as Equation (6): and E is the average energy of adsorption (kJ/mol), which is expressed from β in the following Equation (7): The adsorption isotherms under various temperatures are illustrated in Figure 5a. Table 2 shows relative parameters and correlation coefficients summarized from three adsorption isotherm models. The highest values of correlation coefficients R 2 (>0.9928) were obtained by fitting with the Freundlich model, indicating that the adsorption was taking place on a structurally heterogeneous adsorbent surface. In this study, the value of the heterogeneous factor 1/n is between 0.4667 and 0.4739 (<0.5), which suggests a favorable adsorption in the present research [40]. Further, the maximum adsorption capacity of 172.41 mg/g was obtained from the Langmuir model at 298 K. Based on D-R isotherm, the process of ammonium adsorption could be related to a pore volume filling process [39]. The point of E distinguishes the class of sorption. Thus, physical sorption is determined by the E values in the range from 1 to 8 kJ/mol, while chemical sorption is determined by the E values in the range from 8 to 16 kJ/mol. E values obtained in present study ranged in 8.9087-9.6225 kJ/mol, demonstrating that the adsorption was predominantly chemisorption.
and E is the average energy of adsorption (kJ/mol), which is expressed from β in the following Equation (7): The adsorption isotherms under various temperatures are illustrated in Figure 5a. Table 2 shows relative parameters and correlation coefficients summarized from three adsorption isotherm models. The highest values of correlation coefficients R 2 (>0.9928) were obtained by fitting with the Freundlich model, indicating that the adsorption was taking place on a structurally heterogeneous adsorbent surface. In this study, the value of the heterogeneous factor 1/n is between 0.4667 and 0.4739 (<0.5), which suggests a favorable adsorption in the present research [40]. Further, the maximum adsorption capacity of 172.41 mg/g was obtained from the Langmuir model at 298 K. Based on D-R isotherm, the process of ammonium adsorption could be related to a pore volume filling process [39]. The point of E distinguishes the class of sorption. Thus, physical sorption is determined by the E values in the range from 1 to 8 kJ/mol, while chemical sorption is determined by the E values in the range from 8 to 16 kJ/mol. E values obtained in present study ranged in 8.9087-9.6225 kJ/mol, demonstrating that the adsorption was predominantly chemisorption.   The maximum adsorption capacity in the present study was obviously higher than in other literature. For example, the ammonium exchange capacity by high silica zeolites was 4.08 mg/g [5]. Fu et al. [27] obtained the maximum adsorption amount of ammonium of 16.96 mg/g by using zeolite modified with sodium nitrate (NaNO 3 ). Kamyab and Williams [12] reported that the maximum adsorptive amount of ammonium by Linde Type J zeolite was 51.97 mg/g. Meanwhile, Shabanet al. [41] found that the adsorption capacities of ammonium by clinoptilolite and synthetic zeolite-A were, respectively, 92 mg/g and 99 mg/g. Considering the differences, the main reason may be that the nano magnetic particles enhanced the BET specific surface area and pore volume of the adsorbent (Table 2), which promoted greater adsorption ability for the ammonium. Moreover, Vaičiukynienė et al. [42] stated that the increasing initial ammonium concentration encouraged the internal micropores of the adsorbent to take part in ammonium exchange, which facilitated the great adsorption capacity obtained.

Thermodynamic Parameters
The distribution coefficient, K d , is represented as Equation (8), and the Gibbs free energy, entropy, and enthalpy are calculated by the temperature-dependent adsorption isotherm (Equations (9) and (10)): Figure 5b tabulated the relationship between lnK d and 1/T, while Table 3 summarized the thermodynamic parameters values. The value of ∆G 0 , lower than zero, and the value of ∆H 0 , higher than zero, determined that the adsorption was endothermic, feasible, and spontaneous. The lower values of ∆G 0 coupled with higher temperature (Table 3), determined that higher temperatures promoted the endothermic adsorption. The values of ∆S 0 that were higher than zero revealed that the adsorption was randomness increasing. Table 3. Thermodynamic parameters of ammonium adsorption by M-Zeo.

Adsorption Kinetics
Four typical kinetic models simulated the adsorption kinetics are expressed as follows [43][44][45][46]: Pseudo first-order equation : ln(q e − q t ) = ln q e − k 1 t Pseudo sec ond-order equation : Intraparticle diffusion equation : q t = k 3 t 0.5 (14) here, q t is the amount of adsorption at time t, mg/g; q e is the amount of adsorption at equilibrium, mg/g; k 1 and k 2 are, respectively, the rate constants of pseudo first-order adsorption and pseudo second-order adsorption, g/(mg·h); a e is the initial adsorption rate, mg/(g·h); b e is described to the extent of surface coverage and the activation energy for chemisorption, g/mg; k 3 is the rate constant of intraparticle diffusion, mg/(g·h −0.5 ). Figure 6a shows the adsorption kinetics at pH 8. It is observed that the adsorption significantly rose with the increasing contact time after 1 h. The adsorption quantity was 151.22 mg/g at a contact time of 12 h at 298 K. The data of adsorption kinetics stated that M-Zeo presented a fast adsorption rate for ammonium removal.
Pseudo first-order equation: Intraparticle diffusion equation: here, qt is the amount of adsorption at time t, mg/g; qe is the amount of adsorption at equilibrium, mg/g; k1 and k2 are, respectively, the rate constants of pseudo first-order adsorption and pseudo second-order adsorption, g/(mg·h); ae is the initial adsorption rate, mg/(g·h); be is described to the extent of surface coverage and the activation energy for chemisorption, g/mg; k3 is the rate constant of intraparticle diffusion, mg/(g·h −0.5 ). Figure 6a shows the adsorption kinetics at pH 8. It is observed that the adsorption significantly rose with the increasing contact time after 1 h. The adsorption quantity was 151.22 mg/g at a contact time of 12 h at 298 K. The data of adsorption kinetics stated that M-Zeo presented a fast adsorption rate for ammonium removal. The kinetic parameters after kinetic data in the current study fitted by the above four kinetic models were summarized in Table 4. It is observed that the adsorption kinetic data can be satisfactorily simulated by a pseudo-second-order model with the highest correlation coefficient R 2 =0.9990. According to the pseudo-second-order model, the theoretically adsorbed amount at equilibrium was calculated as 156.25 mg/g, which was close to the value obtained from the experiment (151.22 mg/g). Because of the correlation coefficient R 2 higher than 0.99, the Elovich model acceptably fitted the kinetic data, sup- The kinetic parameters after kinetic data in the current study fitted by the above four kinetic models were summarized in Table 4. It is observed that the adsorption kinetic data can be satisfactorily simulated by a pseudo-second-order model with the highest correlation coefficient R 2 = 0.9990. According to the pseudo-second-order model, the theoretically adsorbed amount at equilibrium was calculated as 156.25 mg/g, which was close to the value obtained from the experiment (151.22 mg/g). Because of the correlation coefficient R 2 higher than 0.99, the Elovich model acceptably fitted the kinetic data, supposing that the adsorption was energetically heterogeneous [47]. As observed in Figure 6b, the adsorption kinetic data can be satisfactorily simulated by the intraparticle diffusion model if the entire curve is separated into two linear patterns. Therefore, the adsorption can be considered to involve two stages, respectively, corresponding to the boundary layer diffusion and the intraparticle diffusion [11]. The fast stage at the first two hours belonged to the rapid occupation of major surface adsorption positions by ammonium, while the gradual stage is ascribed to ammonium entering the internal pores of M-Zeo by the intraparticle diffusion [10]. Thus, the adsorption is firstly rapid adsorption and then gradual equilibrium.

Regeneration
The regeneration of the M-Zeo performance for ammonium adsorption is exhibited in Figure 7. After the second regeneration cycle, the adsorption quantities were clearly increased from 153.49 mg/g to 166.00 mg/g. However, after regeneration by the third cycle, the adsorption quantities were declined. After five regeneration cycles, the adsorption quantity was 148.94 mg/g, with the regeneration ratio of 97.03%. Meanwhile, M-Zeo was proven to have great advantages on magnetic recovery and reused properties by employing the NaCl regeneration method. The ion-exchange equilibrium of the process of "NaCl regeneration" of ammonium-bearing zeolite can be expressed as the following [  The ion-exchange equilibrium of the process of "NaCl regeneration" of ammoniumbearing zeolite can be expressed as the following [48]: On the surface of zeolite, Na + exchanged the position of NH 4 + and then zeolite recovered the property of ammonium exchanging [49]. Meanwhile, Li et al. [49] proved that after three cycles in NaCl regeneration the adsorbent had the adsorption capacity close to that of fresh zeolite. Thus, NaCl solution can act as both a desorbing agent and regenerating agent, which contributes to NH 4 + desorption and M-Zeo regeneration, simultaneously [50].

Conclusions
The M-Zeo with high magnetic recovery performance was successfully synthesized. M-Zeo demonstrated a great performance of ammonium removal from liquid. The Freundlich model and the pseudo-second-order model satisfactorily simulated the adsorption isotherm and kinetics for ammonium adsorption on M-Zeo, respectively. The maximum adsorptive capacity of 172.41 mg/g was obtained. The E values in the range of 8.9087-9.6225 kJ/mol from the D-R model proved that the adsorption process was chemisorption. Thermodynamic parameters determined that the adsorption was endothermic, feasible, and spontaneous. The efficient ammonium removals presented that the economic-efficient, great ammonium affinity, and excellent regeneration characteristics of M-Zeo can be a promising adsorbent extensively utilized in the ammonium treatment of liquid.