Kinetic and isotherm studies of nitrate adsorption on granular Fe–Zr–chitosan complex and electrochemical reduction of nitrate from the spent regenerant solution

Abstract

In this study, a granular Fe–Zr–chitosan complex was synthesized to remove nitrate from aqueous solution and an undivided cylindrical electrochemical cell was constructed to treat the spent regenerant solution, thus achieving separation and conversion. Adsorbent characterization was conducted by SEM, XRD, BET, FTIR and XPS. The effects of pH and co-existing anions on nitrate adsorption were investigated. The results indicated that nitrate adsorption on Fe–Zr–chitosan complex followed the first-order kinetic model and the two-site Langmuir isotherm model. The specific surface area of Fe–Zr–chitosan complex was 23.6 m² g⁻¹ and the maximum adsorption capacity reached 10.60 ± 0.74 mg g⁻¹ (as N). The point of zero charge of Fe–Zr–chitosan complex occurred at a pH of 6.3. No significant changes in nitrate removal efficiency were observed over a wide pH range of 3.0–11.0. Sulphate and phosphate seriously interfered in nitrate removal. The developed electrochemical method might have a potential of practical application for the treatment of the spent regenerant solution.

---

1. Introduction

The degradation of groundwater quality and the contamination of drinking water supplies caused by excess nitrate have received widespread concern. Chemical fertilizers, industrial effluent, domestic wastewater, septic systems and animal manures were primarily responsible for nitrate contamination in natural waters.¹ Anthropogenic input of nitrate to slow-flow water bodies (river, lake, reservoir and bay etc.) adversely affected aquatic plants and animals, accompanying with the development of anoxic zones and the eutrophication of aquatic environments.² Elevated nitrate concentration in drinking water might cause health risks such as methemoglobinemia in infants and stomach cancer in adults.³ Therefore, the fast and effective removal of nitrate was of prime importance for preserving water quality and maintaining sustainable water resources.

Adsorption has recently become an attractive option for contaminant removal due to ease of operation, simplicity of design and cost-effectiveness.⁴ Various adsorbents including surfactant modified zeolites,⁵ kaolinite,⁶ layered double hydroxides⁷ and biochar⁸ have been developed to remove nitrate from aqueous solution. However, these materials had several inherent limitations including small particle size, bad acid–alkali resistance and poor desorption efficiency, which imposed restrictions on their practical application.

Chitosan has recently received considerable attention due to low cost and high contents of amino and hydroxyl functional groups, which imparted it with promising adsorption potential for the removal of various aquatic contaminants.⁹ Several surface modification techniques including protonation, quaternary amine modification and impregnation of metal salts have been employed to increase surface positive charge or provide new surface functional groups with good affinities for nitrate.¹⁰ The appropriate incorporation of metallic oxides into chitosan could contribute to improving the properties of chitosan in virtue of their tunable properties and synergistic natures.¹¹ Chitosan was reported to dissolve in FeCl₃ solution without the addition of other chemical agents due to the protonation of amino groups,¹² whereas zirconia was shown to enhance the properties of chitosan because of its high fracture strength and good acid–alkali resistance.¹³ Recently, Shinde et al. proposed Fe–chitosan and Zr–chitosan complexes for selective arsenic(V) preconcentration.¹⁴ However, the use of granular Fe–Zr–chitosan complex as an adsorbent for the removal of nitrate from aqueous solution has not been reported. Furthermore, adsorption could not convert nitrate into nontoxic nitrogen and regeneration also produced large numbers of secondary brine wastes. Thus, to simultaneously achieve separation and conversion, the electrochemical method was proposed in this study to treat the spent regenerant solution.

The objectives of this work were: (i) to reveal nitrate adsorption mechanisms by characterization analysis; (ii) to analyze kinetic data by new kinetic models; (iii) to predict the maximum adsorption capacity using isotherm models; and (iv) to evaluate the reusability of the adsorbent through desorption studies. Finally, the spent regenerant solution was treated harmlessly through the electrochemical method.

2. Materials and methods

2.1. Adsorbent synthesis

All of the adsorbent synthesis steps were conducted at room temperature (25 ± 1 °C), unless otherwise stated. A desired amount of ferric chloride (FeCl₃·6H₂O) and zirconium oxychloride (ZrOCl₂·8H₂O) was added to 300 mL deionized water (the molar ratio of Fe³⁺ to Zr⁴⁺ 2 : 1), and then continuously stirred until dissolved completely. Subsequently, 9.0 g of chitosan powder was dissolved in the mixed solution (pH = 5.6) for 4 h. The resulting solution was added stepwise to approximately 8% of ammonia solution using a syringe and Fe–Zr–chitosan hydrogel beads formed were allowed to stand for 1.0 h. The beads were separated by decantation, followed by sufficiently washing with deionized water, and then dried at 50 °C for 8 h.

The dried beads were immersed into dilute HCl solution (pH = 2.5, the ratio of solid to liquid = 100 g L⁻¹) at 40 °C and 140 rpm for 4 h in a horizontal shaker. The immersed beads were separated, followed by washing with deionized water, and then dried at 50 °C for 3 h. After two immersions, Granular Fe–Zr–chitosan complex with a particle size range of 1.5–2.0 mm was obtained.

2.2. pH and co-existing anions

50 mL of nitrate solution (50 mg L⁻¹) was poured into a series of conical flasks with 1.0 g of adsorbent, which were sealed and agitated at 140 rpm in a thermostatic shaker at room temperature for 1.5 h. The pH of solution was adjusted using 0.1 mol L⁻¹ HCl or NaOH solution. The concentrations of co-existing anions (Cl⁻, HCO₃⁻, F⁻, SO₄²⁻ and PO₄³⁻) as sodium salts were set at 2 mmol L⁻¹ and 4 mmol L⁻¹.

2.3. Adsorption kinetics

100 mL of nitrate solution (50, 100 and 150 mg L⁻¹) was poured into conical flasks with 2.0 g of adsorbent, which were agitated as mentioned above. 1 mL sample was withdrawn at preset time intervals to determine the residual nitrate concentration.

In this study, the first-order and second-order kinetic models were proposed to accurately reflect the variation of nitrate concentration with time and quantitatively describe adsorption and desorption behaviors at the solid/solution interfaces, which were expressed as:

where C₀ (mg L⁻¹) was the initial nitrate concentration; C_t (mg L⁻¹) was the nitrate concentration at time t; k₁ (min⁻¹) and k₂ (L (mg⁻¹ min⁻¹)) were the adsorption rate constants of the first-order and second-order kinetic models, respectively; k₋₁ (min⁻¹) and k₋₂ (L (mg⁻¹ min⁻¹)) were the corresponding desorption rate constants.

The integrated forms at the boundary condition t = 0 to t = t and C_t = C₀ to C_t = C_t were given as:

The equilibrium concentration of target contaminant in aqueous solution was written as:

2.4. Adsorption isotherm

Different amounts of adsorbent (0.5, 1.0 and 1.5 g) were added to a series of conical flasks containing 50 mL of nitrate solution (30−180 mg L⁻¹). 1 mL sample was taken out after 1.5 h to analyze the residual nitrate concentration under the same operating condition.

In this study, the Langmuir,¹⁵ Freundlich¹⁶ and two-site Langmuir¹⁷ isotherm models were employed to analyze adsorption equilibrium data, which were expressed as:

q_e = K_FC_e^1/n (8)

where q_e (mg g⁻¹) and C_e (mg L⁻¹) were the amount of nitrate uptake and nitrate concentration at equilibrium, respectively; Q_max (mg g⁻¹) was the maximum adsorption capacity; K_L (L mg⁻¹) was the Langmuir constant; K_F ((mg g⁻¹) (L mg⁻¹)^1/n) was the Freundlich constant; n was an empirical parameter; K₁ (L mg⁻¹) and K₂ (L mg⁻¹) were the affinity coefficients; Q₁ (mg g⁻¹) and Q₂ (mg g⁻¹) were the corresponding maximum adsorption capacity (Q_max = Q₁ + Q₂).

2.5. Electrochemical treatment of spent regenerant solution

A Ti/IrO₂–Pt plate (4.5 cm × 15 cm) was used as the anode, whereas a Cu/Zn (Cu: 62 wt%; Zn: 38 wt%) plate with the same dimension was employed as the cathode. The two electrodes with a distance of 10 mm were vertically dipped into an undivided cylindrical electrochemical cell containing 300 mL of the spent regenerant solution under constant stirring with a magnetic stirrer. A DC power supply (PS-305DF, Longwei, China) provided 20 mA cm⁻² of current density. 1.5 mL of sample was taken from the cell at preset time intervals to analyze the concentrations of residual nitrate, nitrite and ammonia (as N) using a UV/vis spectrophotometer (DR 6000, HACH, USA).

2.6. Characterization

The surface morphology and particle size of Fe–Zr–chitosan complex were observed using a scanning electron microscope (SEM) (SSX-550, Shimadzu, Japan). The pH of the point of zero charge (pH_PZC) was obtained by means of pH shift analysis, as reported by Xiao et al.¹⁸ The crystalline structure was analyzed by an X-ray diffractometer (XRD) (D8 Focus, Bruker, Germany) equipped with Cu Kα radiation. The specific surface area, pore volume and pore size distribution were determined by N₂ adsorption/desorption isotherm analysis using an automated gas sorption system (Autosorb-iQ, Quantachrome, USA). FTIR spectra were recorded on a Fourier transform infrared spectrometer (VERTEX 70V, Bruker, Germany) with a resolution of 0.4 cm⁻¹. The surface chemistry of the adsorbents before and after nitrate adsorption was conducted by the X-ray photoelectron spectroscopy (XPS) (ESCALAB 250Xi, Thermo Fisher, USA) with a monochromatic Al Kα X-ray source (1486.6 eV). To compensate for the charging effect, all binding energies were referenced to the neutral carbon peak at 284.6 eV. The software XPS-peak 4.1 was used to fit the XPS spectra peaks and the background subtraction procedure was “Linear”.¹⁹

3. Results and discussion

3.1. Adsorbent characterization

3.1.1. SEM. It is evident that Fe–Zr–chitosan complex has the rough surface, fingerprint-like texture and abundant cracks (Fig. 1a). This surface morphology contributes to increasing specific surface area and thereby providing more active sites for nitrate adsorption. However, the surface of Fe–Zr–chitosan complex after nitrate adsorption is relatively smooth with few cracks (Fig. 1b). This change in surface morphology may be ascribed to the slight swelling of the adsorbent, followed by dying.

Fig. 1 (a) SEM image before nitrate adsorption; (b) SEM image after nitrate adsorption (c) pore size distribution; (d) XRD pattern (A: chitosan, B: before nitrate adsorption, C: after nitrate adsorption).

3.1.2. BET. The specific surface area, total pore volume and average pore diameter of Fe–Zr–chitosan complex are determined to be 23.6 m² g⁻¹ obtained from multipoint BET, 0.012 cm³ g⁻¹ according to BJH method cumulative adsorption pore volume and 2.08 nm, respectively. As illustrated in Fig. 1c, the pore size distribution mainly concentrates in an extremely narrow range of 2–4 nm (2–50 nm, accounting for 71.6%), indicative of a mesoporous material according to the IUPAC classification. Moreover, the pore size distribution exerted a significant effect on the intraparticle diffusion of the target contaminant. In this study, this pore size distribution and abundant cracks contribute to the diffusion of nitrate ions into the interior of Fe–Zr–chitosan complex.

3.1.3. XRD. As shown in Fig. 1d, the XRD pattern of chitosan shows two reflections at 2θ = 14.4° and 19.8°, which correspond to its anhydrous and hydrated crystal structure, respectively.²⁰ However, the peak at 2θ = 19.8° almost disappears for Fe–Zr–chitosan complex, which may be attributed to the fact that the incorporation of Fe³⁺ and Zr⁴⁺ into chitosan destroyed the original ordered structure and reduced the hydrogen bonding ability of chitosan molecules. This behavior was similar to other study where the incorporation of ZnO particles into chitosan impeded the order of polymer chains by both steric effect and inter-molecular hydrogen bonds.²¹ Few differences are observed between XRD patterns for Fe–Zr–chitosan complex before and after nitrate adsorption, indicating that no new crystalline compounds appeared on adsorbent surface after nitrate adsorption.

3.1.4. FTIR. As shown in Fig. 2, the broad and strong band at around 3434 cm⁻¹ corresponds to the superposition of O–H and N–H stretching vibrations of chitosan molecules.²² The other characteristic bands for chitosan are as follows: 2920 cm⁻¹ and 2854 cm⁻¹ (asymmetric and symmetric C–H stretching vibration of –CH₂, respectively), 1649 cm⁻¹ (C=O stretching vibration), 1400 cm⁻¹ (C–O stretching vibration of primary alcohol), 1159 cm⁻¹ (C–O–C stretching vibration), 1084 cm⁻¹ (C–O stretching vibration of secondary alcohol), 895 cm⁻¹ (N–H rocking vibration of –NH₂) and 663 cm⁻¹ (O–H out-of-plane bending vibration).

Fig. 2 FTIR spectra (A: chitosan, B: before nitrate adsorption, C: after nitrate adsorption).

An emerging peak at 458 cm⁻¹ for Fe–Zr–chitosan complex was attributed to Zr–O stretching vibration.¹¹ The red shift from 3434 cm⁻¹ to 3422 cm⁻¹ was attributed to the coordination of Fe³⁺ and Zr⁴⁺ with hydroxyl and amino groups, whereas the peak shifting from 1084 cm⁻¹ to 1076 cm⁻¹ further confirmed hydroxyl groups on C-3 position. There were few changes for FTIR spectra of Fe–Zr–chitosan complex before and after nitrate adsorption, indicating that adsorption mechanism was partially electrostatic attraction.

3.1.5. XPS. The XPS spectra of Fe–Zr–chitosan complex before and after nitrate adsorption were examined to further gain insight into adsorption mechanisms. As shown in Fig. 3a, the XPS survey spectrum clearly confirms the presence of C, N, O, Cl, Fe and Zr. The atomic surface concentration of N element increases from 3.06% to 5.55%, whereas that of O element ranges from 19.86% to 27.76% after nitrate adsorption, suggesting that nitrate ions are successfully absorbed onto the surface of Fe–Zr–chitosan complex.

Fig. 3 (a) XPS survey spectra, (b) C 1s, (c) O 1s and (d) N 1s XPS spectra of Fe–Zr–chitosan complex before and after nitrate adsorption.

The XPS high resolution spectrum of C 1s region of Fe–Zr–chitosan complex (Fig. 3b) can be decomposed into four components centered at 284.4 eV, 285.0 eV assigned to C–H, C–C and at 286.1 eV, 286.8 eV associated with C–O, C–N (Table 1). The O 1s core-level spectrum can be curve-fitted into five individual component peaks (Fig. 3c): 530.1 eV (Fe₂O₃/ZrO₂), 531.2 eV (Fe(OH)₃), 532.1 eV (O–H), 533.1 eV (–C–O–C–) and 533.7 eV (C–OH), respectively. The presence of Fe₂O₃/ZrO₂ and Fe(OH)₃ increases mechanical strength and acid–alkali resistance of Fe–Zr–chitosan complex.¹³ The full width at half maximum and the peak area ratios of C 1s and O 1s have distinct differences after nitrate adsorption, whereas few significant changes are observed in terms of the corresponding binding energies, indicating that the involvement of these functional groups into nitrate adsorption was a physical adsorption.

Table 1 Assignments of main spectral bands based on their binding energies (BEs), full width at half maximum (FWHM) and peak area ratios (PAR) for Fe–Zr–chitosan complex before nitrate adsorption

Elements	BE (eV)	Structure	FWHM	PAR (%)	References
C 1s	284.4	C–H, C–C	1.05	23.41	23 and 24
	285.0	C–H, C–C	1.11	38.77	23 and 24
	286.1	C–O, C–N	1.39	15.42	25
	286.8	C–O, C–N	2.07	22.40	26
N 1s	399.2	–NH^+, –NH2	1.61	42.3	27
	399.7	–NH2	1.49	34.8	28
	400.8	O=C–NH	2.31	22.9	27
O 1s	530.1	Fe2O3, ZrO2	1.25	8.60	29
	531.2	Fe(OH)3	1.15	19.20	29
	532.1	O–H	1.26	21.29	25
	533.1	–C–O–C–	1.10	28.35	30
	533.7	C–OH	1.19	22.56	30

---

As shown in Fig. 3d, a new peak occurs at 407.3 eV for Fe–Zr–chitosan complex after nitrate adsorption, which corresponds to nitrate ions.²⁹ It can be clearly seen from Table 1 that the N 1s core-level spectrum of Fe–Zr–chitosan complex is divided into three peaks at 399.2 eV, 399.7 eV and 400.8 eV, which corresponds to –NH⁺ (–NH₂), –NH₂, O=C–NH, respectively. After nitrate adsorption, the binding energies of the last two peaks of N 1s shift to 400 eV and 401.3 eV, respectively. The results prove that a chemical adsorption may have occurred between chitosan and nitrate ions.³⁰ The protonation of these functional groups in chitosan molecules reduced the electron cloud density of N atom, thus leading to an increase in the binding energies. Therefore, electrostatic attraction may be responsible for the interaction between Fe–Zr–chitosan complex and nitrate ions.

3.2. Effect of pH

The pH of solution was a vital parameter that affected chemical species of solute and surface properties of adsorbent such as surface charge and dissociation of functional groups.³¹ In the present study, the pH of the point of zero charge (pH_pzc) for Fe–Zr–chitosan complex was found to be 6.3. Fig. 4a clearly shows that nitrate removal efficiency (>80%) do not significantly change over a wide pH range of 3.0–11.0, indicating that Fe–Zr–chitosan complex has good acid–alkali resistance. After nitrate adsorption, the occurrence of slight swelling for Fe–Zr–chitosan complex contributes to reducing intraparticle diffusion resistance and enhancing the penetration of nitrate ions into the inner layer of adsorbent. It should be noted that the amount of nitrate uptake for the immersed adsorbent increased by 17.6% due to the swelling effect with an adsorbent dosage of 20 g L⁻¹ and an initial nitrate concentration of 50 mg L⁻¹ under room condition. Additionally, the final pH of solution is located between 4.9 and 5.4, implying that Fe–Zr–chitosan complex has buffer ability.

Fig. 4 Effects of pH (a) and co-existing anions (b) on nitrate adsorption.

Nitrate removal efficiency decreases significantly with further increasing of initial pH and only reaches 63.2% (pH = 11.5) and 14.7% (pH = 12), respectively. These results could be explained as follows: (i) negatively charged adsorbent surface (more than pH_pzc) failed to adsorb nitrate ions due to electrostatic repulsion; (ii) exchangeable chloride ions on adsorbent surface were replaced by hydroxide ions, limiting the exchange between chloride and nitrate ions; and (iii) abundant hydroxide ions increased diffusion resistance of nitrate ions.

3.3. Effect of co-existing anions

The nitrate contaminated water often contains a certain amount of other anions that may compete with nitrate ions for adsorption sites, leading to the decrease in nitrate removal efficiency.³² Compared with the blank experiment, the presence of co-existing anions exerts an adverse influence on nitrate removal and this trend also increases with an increase in their concentrations (Fig. 4b). The adverse effects of these anions on nitrate removal efficiency obey the following order: SO₄²⁻ > PO₄³⁻ > F⁻, HCO₃⁻ > Cl⁻. It was worth noting that accumulation of these co-existing anions on adsorbent surface might promote the formation of negatively charged surface sites, hindering nitrate removal due to electrostatic repulsion.

The pH of solution before and after nitrate adsorption and first-order hydrolysis constant of selected anions are depicted in Table 2. Obviously, the pH of solution increases with an increase in hydrolysis constant under the same anion concentration. No significant changes are observed for Cl⁻, F⁻ and SO₄²⁻ after nitrate adsorption, while the pH of solution decreases significantly for HCO₃⁻ and PO₄³⁻ due to the ion exchange between Cl⁻ on adsorbent surface and OH⁻ produced from HCO₃⁻ and PO₄³⁻ hydrolysis in aqueous solution.

Table 2 The pH of solution before and after nitrate adsorption and hydrolysis constant of co-existing anions

Concentration	Cl^−	SO4^2−	F^−	HCO3^−	PO4^3−
Kd (25 °C)	—	1.0 × 10^−12	2.86 × 10^−11	2.38 × 10^−8	2.27 × 10^−2
2 mmol L^−1	5.86	5.95	6.30	8.41	11.49
	5.77	5.82	5.81	5.93	8.20
4 mmol L^−1	5.95	6.02	6.38	8.49	11.73
	5.83	6.20	6.05	6.49	9.49

---

The adsorption mechanisms of various anions on adsorbent surface have been reported to be primarily electrostatic attraction and/or ion exchange.³³ The pH of solution was below pH_pzc after nitrate adsorption in the presence of Cl⁻, F⁻ and SO₄²⁻, showing the positively charged adsorbent surface. The decrease in nitrate removal efficiency was partially caused by the competition between Cl⁻ (F⁻ and SO₄²⁻) and NO₃⁻ for positively charged adsorption sites. In addition, the ion exchange between SO₄²⁻ and F⁻ in aqueous solution and the exchangeable Cl⁻ on adsorbent surface also led to the decrease in nitrate removal efficiency because SO₄²⁻ could associate with Zr⁴⁺ forming strong surface complex³⁴ whereas Fe³⁺ and Zr⁴⁺ had a good coordination ability for F⁻.

After nitrate adsorption, the pH of solution increases from 5.93 to 6.49 going across pH_pzc when the concentration of HCO₃⁻ ranges from 2 to 4 mM (Table 2), indicating that the sign of charge on adsorbent surface transits from plus to minus. In other words, the sign of charge on adsorbent surface depends on the concentration of HCO₃⁻. Therefore, nitrate adsorption can occur by electrostatic attraction at lower HCO₃⁻ concentration. The pH of solution is beyond pH_pzc in the presence of PO₄³⁻ after nitrate adsorption, indicating that the negatively charged adsorbent surface has an adverse effect on nitrate removal due to electrostatic repulsion.

3.4. Kinetic studies

As shown in Fig. 5a, the concentration of nitrate decreases rapidly in the inception phase due to high concentration gradient at the solid/solution interfaces, followed by a slow decrease process. There are no distinct differences in equilibrium time (around 90 min) under different nitrate concentrations. The amounts of nitrate uptake at equilibrium are calculated to be 2.14, 3.72 and 4.62 mg g⁻¹ for initial nitrate concentrations of 50, 100 and 150 mg L⁻¹, respectively.

Fig. 5 Kinetic (a) and isotherm (b) studies for nitrate adsorption.

The residual sum of squares (RSS),³⁵ chi-square analysis (χ²)³⁶ and coefficient of determination (R²)³⁷ were employed to evaluate the goodness of fit various kinetic and isotherm models, which were defined as follows:

RSS = ∑(q_exp − q_cal)² (10)

where q_exp (mg g⁻¹) was the experimental nitrate uptake at time t; q_cal (mg g⁻¹) was the calculated nitrate uptake; and q̄ (mg g⁻¹) was the average value of q_exp.

Mathematically speaking, the smaller RSS value, the closer the data points are to the fitted curve. The R² is used as one measure of the quality of the regression models and a value of R² close to 1 indicates that the fit is a good one. When performing nonlinear curve fitting, an interactive procedure that minimizes the value of χ² is employed to obtain the optimal parameter values.

As shown in Table 3, compared with the second-order kinetic model, the calculated values (C_cal) obtained from the first-order kinetic model are closer to the experimental values (C_exp) under different nitrate concentrations when adsorption equilibrium is reached. The first-order kinetic model also provides the smaller RSS and χ² values and higher R² value. In addition, the fitted curve given by the first-order kinetic model is closer to the experimental data points (Fig. 5a). These results indicated that the adsorption of nitrate on granular Fe–Zr–chitosan complex followed the first-order kinetic model.

Table 3 Kinetic parameters obtained from the first-order and second-order kinetic models

C0 (mg L^−1)	First-order kinetic model				RSS	χ^2	R^2
	Cexp (mg L^−1)	Ccal (mg L^−1)	k1 (min^−1)	k−1 (min^−1)
50	6.71	6.23	(6.21 ± 0.10) × 10^−2	(8.93 ± 0.49) × 10^−3	1.63	0.20	0.9992
100	26.65	25.83	(4.80 ± 0.04) × 10^−2	(1.65 ± 0.03) × 10^−2	1.64	0.20	0.9997
150	55.31	53.22	(3.32 ± 0.05) × 10^−2	(1.87 ± 0.05) × 10^−2	6.79	0.85	0.9993

---

C0 (mg L^−1)	Second-order kinetic model				RSS	χ^2	R^2
	Cexp (mg L^−1)	Ccal (mg L^−1)	k2 (g (mg^−1 min^−1))	k−2 (g (mg^−1 min^−1))
50	6.71	5.05	(1.86 ± 0.12) × 10^−3	(2.38 ± 0.47) × 10^−5	25.46	3.18	0.9871
100	26.65	23.39	(5.83 ± 0.22) × 10^−4	(5.29 ± 1.04) × 10^−5	25.03	3.13	0.9959
150	55.31	51.28	(2.49 ± 0.05) × 10^−4	(7.03 ± 0.57) × 10^−5	13.80	1.73	0.9986

---

The adsorption rate constant decreases with an increase in initial nitrate concentration, while the desorption rate constant increases (Table 3). In other words, the rate constant suffered from solute concentration, which differed from that of pure chemical reaction. Consequently, nitrate adsorption on granular Fe–Zr–chitosan complex involved different adsorption mechanisms. The adsorption and desorption rate constants provided by the first-order and second-order kinetic models quantitatively reflected the extent of adsorption and desorption reaction at the solid/solution interfaces. The magnitude of the ratio of k₁ to k₋₁ (k₂ to k₋₂) reflected the extent of adsorption reaction. Obviously, adsorption reaction occurred thoroughly to a large extent at low nitrate concentration.

3.5. Equilibrium studies

The effect of adsorbent dosage on nitrate adsorption is depicted in Fig. 5b. It is evident that the amount of nitrate uptake decreases with an increase in adsorbent dosage due to the unsaturation of the available adsorption sites. As shown in Table 4, the Freundlich model is more suitable to describe nitrate adsorption compared with the Langmuir model due to the smaller RSS and χ² values and larger R² value, indicating multilayer adsorption on the heterogeneous surface.¹⁵ It was believed that the adsorption was easy to proceed when the value of n was in a range of 2–10.¹⁶ It is obvious that all of the values of n confirm the favorable conditions for nitrate adsorption (Table 4). However, the Freundlich model as an empirical equation did not predict the maximum adsorption capacity.

Table 4 Isotherm parameters obtained from the Langmuir, Freundlich and Two-site Langmuir models

Models	Parameters	Adsorbent dosage (g L^−1)
		10	20	30
Langmuir	Qmax (mg g^−1)	7.98 ± 0.54	5.95 ± 0.60	4.83 ± 0.60
	KL (L mg^−1)	0.089 ± 0.027	0.087 ± 0.033	0.089 ± 0.037
	RSS	1.18	0.88	0.62
	χ^2	0.30	0.22	0.154
	R^2	0.9244	0.9175	0.9133
Freundlich	KF ((mg g^−1) (L mg^−1)^1/n)	1.93 ± 0.05	1.24 ± 0.04	0.95 ± 0.05
	n	3.38 ± 0.08	2.91 ± 0.07	2.71 ± 0.10
	RSS	2.14 × 10^−2	1.41 × 10^−2	2.35 × 10^−2
	χ^2	5.35 × 10^−3	3.53 × 10^−3	5.88 × 10^−3
	R^2	0.9986	0.9987	0.9967
Two-site Langmuir	Qmax (mg g^−1)	10.60 ± 0.74	8.48 ± 0.39	7.3 ± 0.24
	Q1 (mg g^−1)	7.47 ± 0.27	6.60 ± 0.25	5.97 ± 0.19
	Q2 (mg g^−1)	3.13 ± 0.47	1.88 ± 0.14	1.33 ± 0.05
	K1 (L mg^−1)	0.015 ± 0.004	0.016 ± 0.002	0.017 ± 0.001
	K2 (L mg^−1)	0.96 ± 0.58	1.59 ± 0.44	2.72 ± 0.44
	RSS	6.38 × 10^−3	1.83 × 10^−3	4.39 × 10^−4
	χ^2	3.19 × 10^−3	9.15 × 10^−4	2.20 × 10^−4
	R^2	0.9996	0.9998	0.9999

---

The two-site Langmuir model is most suitable for describing nitrate adsorption due to the lowest RSS and χ² values and highest R² value under different adsorbent dosages (Table 4), suggesting that there might exist two types of active sites with different binding energy on the adsorbent surface.¹⁷ The maximum adsorption capacity predicted by the two-site Langmuir model reaches 10.60 ± 0.74 mg g⁻¹ (as N). Q₁ > Q₂ and K₁ < K₂ indicates that nitrate adsorption mainly occurs on active sites with lower binding energy. The maximum adsorption capacity increased with the decrease in adsorbent dosage because adsorption sites with lower binding energy were more sufficiently occupied under lower dosage. Therefore, adsorption equilibrium moved towards favorable adsorption. It can be clearly seen from Table 5 that Fe–Zr–chitosan complex has a moderate adsorption capacity for nitrate removal compared with other adsorbents. Furthermore, Fe–Zr–chitosan complex has a relatively large particle size and short equilibrium time, which enhance a potential of practical application.

Table 5 A comparison of various adsorbents properties for nitrate removal

Adsorbents	Qmax (mg g^−1)	Particle size	Equilibrium time	References
Agricultural residue	16.8	<250 μm	24 h	38
Modified wheat residue	29.12	150–250 μm	2 h	39
Surfactant loaded zeolites	0.78	1–2 mm	24 h	5
Hydroxyapatite	48.8	<0.55 mm	40 min	40
Fe–Zr–chitosan complex	10.6	1.5–2.0 mm	1.5 h	This study
LDHs (Mg/Al 3 : 1)	8.97	0.2–0.5 mm	30 min	41
MCM-48	1.40	Powder	24 h	42
Hydrous bismuth oxide	0.97	40–50 μm	6 h	43

---

3.6. Regeneration

The feasibility of Fe–Zr–chitosan complex as an adsorbent depends not only on high adsorption capacity, but also on good desorption performance. In the present study, NaCl, Na₂SO₄ and NaOH solutions (50 mmol L⁻¹) were selected as the desorption agents and the corresponding desorption efficiencies were found to be 88.4%, 92.3% and 97% respectively. These results indicated that Fe–Zr–chitosan complex could be effectively regenerated. The amounts of nitrate uptake reached 4.65, 0.79 and 3.92 mg g⁻¹ for NaCl, Na₂SO₄ and NaOH solutions, respectively when the regenerated adsorbent was first used to adsorb nitrate. Zr⁴⁺ and SO₄²⁻ formed strong surface complex during regeneration, providing more negatively charged adsorption sites and thereby leading to the significant decrease in nitrate removal efficiency.³⁴ Thus, the pretreatment is required to promote nitrate removal efficiency when wastewater contains high concentration of sulphate.

3.7. Electrochemical treatment

It is observed from Fig. 6a that the concentration of nitrate decreases gradually with time, reaching 7.28 mg L⁻¹ at the end of electrolysis, while few nitrite and ammonia are detected. Nitrite as an intermediate product was extremely unstable in the electrochemical cell and was readily converted into other nitrogenous compounds. High concentration of chloride ions (Cl⁻) in the spent regenerant solution regenerated by NaCl solution was oxidized to free chlorine (Cl₂) at the anode, followed by the disproportionation reaction of Cl₂ that produced hypochlorous acid (HOCl) in acidic solution or hypochlorite (OCl⁻) in alkaline solution.⁴⁴ It has been reported that HOCl and OCl⁻ could decompose ammonia to nitrogen gas due to their high oxidative potentials.⁴⁵

2NH₃ + 3HOCl → N₂ + 3H⁺ + 3Cl⁻ + 3H₂O (13)

2NH₃ + 3OCl⁻ → N₂ + 3Cl⁻ + 3H₂O (14)

Fig. 6 Electrochemical treatment of spent regeneration solution ((a) NaCl, (b) Na₂SO₄).

As depicted in Fig. 6b, nitrate and nitrite concentrations in the spent regenerant solution regenerated by Na₂SO₄ solution reached 7.69 and 0.4 mg L⁻¹ respectively at the end of electrolysis. However, the concentration of ammonia was very high, reaching 24.1 mg L⁻¹, which was attributed to the following facts that: (i) the direct oxidation at the anode/solution interfaces by the stepwise dehydrogenation of ammonia and the indirect oxidation by hydroxyl radicals could be neglected⁴⁶ and (ii) sulphate ions (SO₄²⁻) in the spent regenerant solution were hardly converted to persulphate ions (S₂O₈²⁻) and thereby had no oxidation ability for ammonia.

It has been reported that nitrate reduction on the cathode occurred through a consecutive reaction mechanism⁴⁷ in which nitrate was first converted to nitrite, followed by the reduction of nitrite to other nitrogenous compounds. A typical reaction pathway could be described as:

Provided that the reduction rates of the above two conversions in the consecutive reaction mechanism followed first order kinetics, and then the corresponding differential equations were expressed as follows:

The solutions of the differential equations were given as:

[NO₃⁻]_t = [NO₃⁻]_i exp(−k₁t) (18)

The corresponding time was calculated by the following expression when the concentration of an intermediate product (i.e. nitrite) reached the maximum value.

The nonlinear regression of the experimental results for the spent regenerant solution regenerated by Na₂SO₄ solution gave k₁ = 1.08 × 10⁻² min⁻¹ and k₂ = 0.26 min⁻¹ according to the above equations, while the time required for the concentration of nitrite reaching the maximum value (1.97 mg L⁻¹) was approximately 13 min. The value of k₂ was 24 times more than that of k₁, indicating that the reduction of nitrate to nitrite was the rate determining step and that the concentration of nitrite was not very high due to the high reduction rate of nitrite to ammonia.

4. Conclusions

Granular Fe–Zr–chitosan complex with high mechanical strength and good acid–alkali resistance exhibited good performance for nitrate adsorption. It could quickly and efficiently remove nitrate from aqueous solution by electrostatic attraction and intraparticle diffusion. The equilibrium time and maximum adsorption capacity were 1.5 h and 10.60 ± 0.74 mg g⁻¹ (as N), respectively. The adverse effects of selected anions on nitrate adsorption obeyed the following order: SO₄²⁻ > PO₄³⁻ > F⁻, HCO₃⁻ > Cl⁻. The adsorption of nitrate on Fe–Zr–chitosan complex agreed well with the first-order kinetic model. Isotherm studies indicated that nitrate adsorption occurred on two types of active sites with different binding energies. Additionally, this adsorbent using NaCl solution as the eluent could be effectively regenerated. The electrochemical method could effectively convert nitrate into nitrogen in the spent regenerant solution and the reduction of nitrate to nitrite was the rate-determining step.

Acknowledgements

The authors acknowledge financial supports from the National Natural Science Foundation of China (NSFC) (No. 51578519; No. 21407129).

References

1. Z. Jin, Z. Pan, M. Jin, F. Li, Y. Wan and B. Gu, Determination of nitrate contamination sources using isotopic and chemical indicators in an agricultural region in China, Agric., Ecosyst. Environ., 2012, 155, 78–86.
2. I. García-Garizábal, J. Causapé and R. Abrahao, Nitrate contamination and its relationship with flood irrigation management, J. Hydrol., 2012, 442–443, 15–22.
3. L. Hajhamad and M. N. Almasri, Assessment of nitrate contamination of groundwater using lumped-parameter models, Environ. Model. Software, 2009, 24, 1073–1087.
4. R. Katal, M. S. Baei, H. T. Rahmati and H. Esfandian, Kinetic, isotherm and thermodynamic study of nitrate adsorption from aqueous solution using modified rice husk, J. Ind. Eng. Chem., 2012, 18, 295–302.
5. H. Guan, E. Bestland, C. Zhu, H. Zhu, D. Albertsdottir, J. Hutson, C. T. Simmons, M. Ginic-Markovic, X. Tao and A. V. Ellis, Variation in performance of surfactant loading and resulting nitrate removal among four selected natural zeolites, J. Hazard. Mater., 2010, 183, 616–621.
6. L. Tribe, R. Hinrichs and J. D. Kubicki, Adsorption of Nitrate on Kaolinite Surfaces: A Theoretical Study, J. Phys. Chem. B, 2012, 116, 11266–11273.
7. A. Halajnia, S. Oustan, N. Najafi, A. R. Khataee and A. Lakzian, Adsorption–desorption characteristics of nitrate, phosphate and sulfate on Mg–Al layered double hydroxide, Appl. Clay Sci., 2013, 80–81, 305–312.
8. Z. Wang, H. Guo, F. Shen, G. Yang, Y. Zhang, Y. Zeng, L. Wang, H. Xiao and S. Deng, Biochar produced from oak sawdust by Lanthanum (La)-involved pyrolysis for adsorption of ammonium (NH₄⁺), nitrate (NO₃⁻), and phosphate (PO₄³⁻), Chemosphere, 2015, 119, 646–653.
9. F. G. L. M. Borsagli, A. A. P. Mansur, P. Chagas, L. C. A. Oliveira and H. S. Mansur, O-Carboxymethyl functionalization of chitosan: complexation and adsorption of Cd(II) and Cr(VI) as heavy metal pollutant ions, React. Funct. Polym., 2015, 97, 37–47.
10. A. Bhatnagar and M. Sillanpää, A review of emerging adsorbents for nitrate removal from water, Chem. Eng. J., 2011, 168, 493–504.
11. H. Jiang, P. Chen, S. Luo, X. Tu, Q. Cao and M. Shu, Synthesis of novel nanocomposite Fe₃O₄/ZrO₂/chitosan and its application for removal of nitrate and phosphate, Appl. Surf. Sci., 2013, 284, 942–949.
12. Q. Hu, N. Chen, C. Feng and W. Hu, Nitrate adsorption from aqueous solution using granular chitosan–Fe³⁺ complex, Appl. Surf. Sci., 2015, 347, 1–9.
13. L. Kljajević, B. Matović, A. Radosavljević-Mihajlović, M. Rosić, S. Bosković and A. Devečerski, Preparation of ZrO₂ and ZrO₂/SiC powders by carbothermal reduction of ZrSiO₄, J. Alloys Compd., 2011, 509, 2203–2215.
14. R. N. Shinde, A. K. Pandey, R. Acharya, R. Guin, S. K. Das, N. S. Rajurkar and P. K. Pujari, Chitosan-transition metal ions complexes for selective arsenic(V) preconcentration, Water Res., 2013, 47, 3497–3506.
15. K. Vijayaraghavan, T. V. N. Padmesh, K. Palanivelu and M. Velan, Biosorption of nickel(II) ions onto Sargassum wightii: application of two-parameter and three-parameter isotherm models, J. Hazard. Mater., 2006, 133, 304–308.
16. M. Jaina, V. K. Garg and K. Kadirvelu, Adsorption of hexavalent chromium from aqueous medium onto carbonaceous adsorbents prepared from waste biomass, J. Environ. Manag., 2010, 91, 949–957.
17. L. Chai, Y. Wang, N. Zhao, W. Yang and X. You, Sulfate-doped Fe₃O₄/Al₂O₃ nanoparticles as a novel adsorbent for fluoride removal from drinking water, Water Res., 2013, 47, 4040–4049.
18. L. Xiao, W. Ma, M. Han and Z. Cheng, The influence of ferric iron in calcined nano-Mg/Al hydrotalcite on adsorption of Cr(VI) from aqueous solution, J. Hazard. Mater., 2011, 186, 690–698.
19. W. Jiang, X. Chen, B. Pan, Q. Zhang, L. Teng, Y. Chen and L. Liu, Spherical polystyrene-supported chitosan thin film of fast kinetics and high capacity for copper removal, J. Hazard. Mater., 2014, 276, 295–301.
20. O. Inta, R. Yoksan and J. Limtrakul, Hydrophobically modified chitosan: a bio-based material for antimicrobial active film, Mater. Sci. Eng., C, 2014, 42, 569–577.
21. S. John, A. Joseph, A. J. Jose and B. Narayana, Enhancement of corrosion protection of mild steel by chitosan/ZnO nanoparticle composite membranes, Prog. Org. Coat., 2015, 84, 28–34.
22. X. Weng, S. Lin, Y. Zhong and Z. Chen, Chitosan stabilized bimetallic Fe/Ni nanoparticles used to remove mixed contaminants-amoxicillin and Cd(II) from aqueous solutions, Chem. Eng. J., 2013, 229, 27–34.
23. L. Huang, S. Yuan, L. Lv, G. Tan, B. Liang and S. O. Pehkonen, Poly(methacrylic acid)-grafted chitosan microspheres via surface initiated ATRP for enhanced removal of Cd(II) ions from aqueous solution, J. Colloid Interface Sci., 2013, 405, 171–182.
24. X. Zhang and R. Bai, Mechanisms and kinetics of humic acid adsorption onto chitosan-coated granules, J. Colloid Interface Sci., 2003, 264, 30–38.
25. J. He, Y. Lu and G. Luo, Ca(II) imprinted chitosan microspheres: an effective and green adsorbent for the removal of Cu(II), Cd(II) and Pb(II) from aqueous solutions, Chem. Eng. J., 2014, 244, 202–208.
26. J. F. B. Barata, R. J. B. Pinto, V. I. R. C. Vaz Serra, A. J. D. Silvestre, T. Trindade, M. G. P. M. S. Neves, J. A. S. Cavaleiro, S. Daina, P. Sadocco and C. S. R. Freire, Fluorescent Bioactive Corrole Grafted-Chitosan Films, Biomacromolecules, 2016, 17, 1395–1403.
27. L. Min, Z. Yuan, L. Zhong, Q. Liu, R. Wu and Y. Zheng, Preparation of chitosan based electrospun nanofiber membrane and its adsorptive removal of arsenate from aqueous solution, Chem. Eng. J., 2015, 267, 132–141.
28. H. Liu, F. Yang, Y. Zheng, J. Kang, J. Qu and J. P. Chen, Improvement of metal adsorption onto chitosan/Sargassum sp. composite sorbent by an innovative ion-imprint technology, Water Res., 2011, 45, 145–154.
29. A. V. Naumkin, A. Kraut-Vass, S. W. Gaarenstroom and C. J. Powell, NIST X-ray Photoelectron Spectroscopy Database, NIST Standard Reference Database 20, Version 4.1, 2000, http://srdata.nist.gov/xps/.
30. P. Huang, M. Cao and Q. Liu, Adsorption of chitosan on chalcopyrite and galena from aqueous suspensions, Colloids Surf., A, 2012, 409, 167–175.
31. D. Kołodynska, Chitosan as an effective low-cost sorbent of heavy metal complexes with the polyaspartic acid, Chem. Eng. J., 2011, 173, 520–529.
32. M. Islam and R. Patel, Synthesis and physicochemical characterization of Zn/Al chloride layered double hydroxide and evaluation of its nitrate removal efficiency, Desalination, 2010, 256, 120–128.
33. P. Ganesan, R. Kamaraj and S. Vasudevan, Application of isotherm, kinetic and thermodynamic models for the adsorption of nitrate ions on graphene from aqueous solution, J. Taiwan Inst. Chem. Eng., 2013, 44, 808–814.
34. X. Liu and L. Zhang, Removal of phosphate anions using the modified chitosan beads: adsorption kinetic, isotherm and mechanism studies, Powder Technol., 2015, 277, 112–119.
35. S. Álvarez-Torrellas, R. S. Ribeiro, H. T. Gomes, G. Ovejero and J. García, Removal of antibiotic compounds by adsorption using glycerol-based carbon materials, Chem. Eng. J., 2016, 296, 277–288.
36. G. McKay, A. Mesdaghinia, S. Nasseri, M. Hadi and M. S. Aminabad, Optimum isotherms of dyes sorption by activated carbon: fractional theoretical capacity & error analysis, Chem. Eng. J., 2014, 251, 236–247.
37. Y. S. Ho, Second-order kinetic model for the sorption of cadmium onto tree fern: a comparison of linear and non-linear methods, Water Res., 2006, 40, 119–125.
38. U. S. Orlando, A. U. Baes, W. Nishijima and M. Okada, Preparation of agricultural residue anion exchangers and its nitrate maximum adsorption capacity, Chemosphere, 2002, 48, 1041–1046.
39. Y. Wang, B. Gao, W. Yue and Q. Yue, Adsorption kinetics of nitrate from aqueous solutions onto modified wheat residue, Colloids Surf., A, 2007, 308, 1–5.
40. M. Islam, P. C. Mishra and R. Patel, Physicochemical characterization of hydroxyapatite and its application towards removal of nitrate from water, J. Environ. Manag., 2010, 91, 1883–1891.
41. A. Halajnia, S. Oustan, N. Najafi, A. R. Khataee and A. Lakzian, The adsorption characteristics of nitrate on Mg–Fe and Mg–Al layered double hydroxides in a simulated soil solution, Appl. Clay Sci., 2012, 70, 28–36.
42. M. K. Seliem, S. Komarneni, T. Byrne, F. S. Cannon, M. G. Shahien, A. A. Khalil and I. M. Abd El-Gaid, Removal of nitrate by synthetic organosilicas and organoclay: kinetic and isotherm studies, Sep. Purif. Technol., 2013, 110, 181–187.
43. P. K. Singh, S. Banerjee, A. L. Srivastava and Y. C. Sharma, Kinetic and equilibrium modeling for removal of nitrate from aqueous solutions and drinking water by a potential adsorbent, hydrous bismuth oxide, RSC Adv., 2015, 5, 35365–35376.
44. D. Reyter, D. Bélanger and L. Roué, Nitrate removal by a paired electrolysis on copper and Ti/IrO₂ coupled electrodes – influence of the anode/cathode surface area ratio, Water Res., 2010, 44, 1918–1926.
45. K. W. Kim, Y. J. Kim, I. T. Kim, G. I. Park and E. H. Lee, The electrolytic decomposition mechanism of ammonia to nitrogen at an IrO₂ anode, Electrochim. Acta, 2005, 50, 4356–4364.
46. L. Li and Y. Liu, Ammonia removal in electrochemical oxidation: mechanism and pseudo-kinetics, J. Hazard. Mater., 2009, 161, 1010–1016.
47. I. Katsounaros, D. Ipsakis, C. Polatides and G. Kyriacou, Efficient electrochemical reduction of nitrate to nitrogen on tin cathode at very high cathodic potentials, Electrochim. Acta, 2006, 52, 1329–1338.

---