Porous graphene oxide based inverse spinel nickel ferrite nanocomposites for the enhanced adsorption removal of arsenic

Abstract

Porous graphene oxide based magnetic inverse spinel nickel ferrite nanocomposites, namely graphene oxide based inverse spinel nickel ferrite (GONF) and reduced graphene oxide based inverse spinel nickel ferrite (rGONF), with particle sizes of around 30 to 40 nm were prepared by the co-precipitation of graphene oxide (GO) with nickel and iron salts in one pot. GONF and rGONF, having ferromagnetic and superparamagnetic properties respectively, were separated easily within 10 seconds, using a small external magnetic field. These nanocomposites were used for the adsorption removal of As(III) and As(V). Compared to bare nickel ferrite, other nanocomposites and GO, the nanocomposites show a high adsorption capacity for As(III) and As(V), with considerable enhancement. The enhanced high adsorption capacity is due to the increased number of pores and adsorption sites with increasing surface area in the GONF and rGONF composites, through reducing the aggregation of bare ferrites. The adsorption results found that more than 99.9% arsenic removal was achieved with the present nanocomposites. Since the nanocomposites show good stability without loss of their adsorption capacity for up to 5 cycles, they can be used for the practical removal of arsenic from water.

---

1. Introduction

Arsenic is a well-known toxic and carcinogenic element which has multiple oxidations states (+5, +3, 0 and −3). Among its several oxidation states, As(III) and As(V) are the most common in natural water or groundwater.¹ Many industrial activities, natural weathering processes, geochemical reactions, biological activity, smelting of metal ores, leaching of man-made arsenic compounds such as desiccants, wood preservatives and agricultural pesticides, and many other anthropogenic activities are the main sources of the release of arsenic into water.^2,3 This means that arsenic contamination arises from both anthropogenic sources and natural geogenic processes. Under extreme conditions, anthropogenic activities may lead to the geogenic release of arsenic into groundwater in a soluble state. A typical example is the Ganges–Brahmaputra–Meghna (GBM) delta in Bangladesh and India, where As contamination in groundwater occurs through a geogenic process which has been the focus of substantial investigation.⁴ The speciation and mobility of As in groundwater are sensitive to both the redox conditions and pH values of the aqueous phase. Generally, aquifers with elevated groundwater exist under reducing geochemical conditions. This means that As(III) is more predominant than As(V) in groundwater. Therefore, it is reasonable to expect that groundwater in other areas that have a similar combination of these geological, geochemical and hydrological conditions may be at risk.⁵ The maximum permissible limit of arsenic in ground water has set to be 0.01 ppm (World Health Organization’s (WHO, Geneva, Switzerland), Environmental Protection Agency (US-EPA, United States, 1993) and 0.05 ppm (Central Pollution Control Board (CPCB, India) standards).^6–8 Drinking of groundwater contaminated with arsenic over a period of time leads to serious arsenic poisoning.⁹ Arsenic poisoning causes several diseases such as skin, lung and kidney cancer, gastrointestinal disease, bone marrow disorder and cardiovascular disease.^10,11 As(III) and As(V) ions that exist in natural water are very toxic. Therefore, generally, adsorbents that have a high adsorption capacity have been applied to remove arsenic ions from water. Widely used adsorbents are zero-valent iron,^12,13 zirconium dioxide impregnated granular activated carbon,¹⁴ copper oxide impregnated alumina,¹⁵ magnetic binary oxide particles,¹⁶ iron hydroxides and oxides,^17–19 and iron-based materials.^20,21 Moreover, iron oxide-based nanomaterials are very effective for the adsorption removal of As(III) and As(V).^9,21–23 However, the practical application of these adsorbents is difficult, due to their small particle size and instability when exposed to the atmosphere.²⁴ To avoid this shortfall, iron oxides have been combined with carbon and carbon nanotubes.^25,26

Many researchers have prepared graphene or graphene oxide (GO) based hybrid nanocomposites for various potential applications.^27–30 Recently, a Fe₃O₄/GO composite,^31,32 and magnetic reduced GO^9,33 have been used for the treatment of arsenic and dyes. Mn₃O₄/GO^34,35 and other GO based hybrid nanocomposites^36,37 have also been used for the adsorption of various organic and inorganic pollutants from water. Recently, nickel ferrite–GO composites have been studied as a promising reacting media, because Ni²⁺ in nickel ferrite exhibits unique properties, such as high catalytic efficiency with high charge (electron) transfer capacity compared to iron ferrites.^38,39 A survey of the literature and the stability of GO based nanocomposites prompted us to synthesize graphene oxide and reduced graphene oxide based inverse spinel nickel ferrite nanocomposites for arsenic removal from aqueous solutions, thereby reducing the potential risks to human health and the environment.

2. Materials and methods

2.1 Preparation of GONF and rGONF nanocomposites

Graphene oxide (GO), prepared from the oxidation of graphite flakes using the modified Hummers' method,⁴⁰ was used for the synthesis of GO based spinel nickel ferrite (GONF),³⁸ and reduced GO based nickel ferrite (rGONF). A pictorial representation of the preparation of GONF and rGONF is shown in Fig. 1. First, 0.5 g of GO was dispersed in 500 mL water by agitating for 1 h with ultra-sonication to transform a portion of the carboxylic acid groups to carboxylate anions. Then, 2.50 g of Fe(NO₃)₃·9H₂O and 0.875 g of Ni(NO₃)₃·9H₂O (2 : 1 molar ratio of Fe³⁺ : Ni²⁺) were dissolved in 15 mL of double distilled water. This solution was then added drop-wise to the GO solution at room temperature under vigorous stirring and nitrogen purging (40 mL min⁻¹). After completing the ion exchange, the solution pH was adjusted to pH ≥ 12.0 by the drop-wise (with dropping rate of 10 drops/min) addition of 28% NH₃ solution under constant magnetic stirring and monitoring of the solution pH using a pH meter. After adjusting the pH, the solution was kept at 80 °C for 45 min under constant stirring, causing a black precipitate to form in the solution. The solution was then cooled to room temperature, and the precipitate was washed several times with de-ionized double distilled water, separated by an external magnetic field, and dried at 60 °C under vacuum for 12 h. The sample obtained was denoted GONF. rGONF was prepared following a similar procedure to the one used for the preparation of GONF. For the preparation of rGONF (Fig. 1), hydrazine hydrate (10 mL) was slowly added to a solution containing 0.5 g GO, and a 2 : 1 molar ratio of Fe³⁺ : Ni² (2.5 g of Fe(NO₃)₃·9H₂O and 0.875 g of Ni(NO₃)₃·9H₂O) at pH ≥ 12 and under a nitrogen atmosphere, for the reduction of GO. The solution temperature was then raised to 80 °C under constant stirring, resulting in a black precipitate forming in the solution. After rapid stirring for 5 h, the solution was cooled down to room temperature. The resulting black precipitate was thoroughly washed with double distilled de-ionized water/ethanol, collected by magnetic separation, and then dried at 60 °C under vacuum for 12 h. This sample was denoted rGONF.

Fig. 1 Pictorial description of the preparation of GO based porous inverse spinel nickel ferrite nanocomposites: GONF and rGONF.

2.2 Adsorption studies of As(III) and As(V) onto GONF and rGONF

A stock solution (1000 mg L⁻¹) of both As(III) and As(V) was prepared from NaAsO₂ and Na₂HAsO₄·7H₂O, respectively. A typical batch adsorption experiment was carried out with a 50 mL aqueous solution of arsenic at pH 6.5 for both As(III) and As(V) in a falcon tube at room temperature. 0.3 g L⁻¹ of GONF or rGONF was added to this solution. Then, the mixture was shaken for pre-determined equilibrium time intervals (150 min), and the samples were filtered out through a 0.45 μm membrane syringe filter for measuring the arsenic concentration using an Optima 2100 DV inductively coupled plasma-optical emission spectrometer (ICP-OES) (PerkinElmer, USA). The reagent blank (arsenic ions without nanocomposites) was checked for measuring the adsorption caused by the falcon tube. The amount of As(III) and As(V) adsorbed onto GONF and rGONF was calculated from the mass balance between the initial (C_o) and equilibrium (C_e) concentrations (mg L⁻¹) of arsenic ions. The arsenic uptake capacity (q_e, mg g⁻¹) of GONF was calculated using this equation: q_e = (C_o − C_e) × Vm⁻¹, where V is the volume of the arsenic solution and m (g) is the weight of GONF. All pH measurements were carried out using a 340i pH meter (WTW, Germany). The adsorption experiment was also performed at different pH values by adjusting the pH with 0.1 M HCl and NaOH solutions. Desorption studies were conducted using a 0.2 M NaOH solution by using 0.5 g L⁻¹ of arsenic ion loaded nanocomposites. All experiments were carried out in triplicate to avoid discrepancies in the experimental results and the values reported are mean ± standard deviation (SD). Origin 8.0© was used to fit the kinetics and equilibrium models using linear regression analysis.

2.3 Preparation of real groundwater spiked with arsenic

Real groundwater was collected from Hagye-dong, Nowon-gu, Seoul, Korea and its physical and chemical compositions are reported in Table S1.† 10 mL of this real water was placed into a 1 L standard flask and diluted to 1 L with double distilled water. To this, 2.0, 4.0 and 8.0 mg L⁻¹ of As(III) or As(V) was added to make three samples, denoted as sample 1, sample 2 and sample 3. These three samples were then used in experiments to test the adsorption removal of arsenic, using 0.3 g L⁻¹ of nanocomposites.

3. Results and discussion

3.1 Characterization of the prepared GONF and rGONF nanocomposites

X-ray diffraction (XRD) results (Fig. S1a†) show that the crystalline properties of GO decrease as the meso-porous carbonaceous nature increases in modified GONF and rGONF. The strong peak at 2θ = 11.3° confirms the formation of GO. Meanwhile, the corresponding peak of GO is shifted to a lower range (2θ = 8.9° and 5.6°) in GONF and rGONF.^41,42 The diffraction peaks at 2θ = 26.6°, 35.65° and 44.5° indicate the presence of spinel nickel ferrite in GONF and rGONF.^39,43 However, the nickel ferrite diffraction peaks (2θ: 26.35°, 35.60° and 43.20°) of rGONF are shifted to a lower range compared to those for GONF. The shifting of the diffraction peaks may be associated with the decreased size and increased porous surface area of GONF due to its reduction. The size of GONF and rGONF was calculated from the well-known Scherrer formula: D = 0.94λ/β cos θ, where D is the crystallite size (nm), β is the full width of the diffraction line at half of the maximum intensity measured in radians, λ is the X-ray wavelength of Cu Kα = 0.154 nm, and θ is the Bragg angle. The decreasing size of GONF from 41.0 nm to 32.12 nm (rGONF) by the reduction of GONF with hydrazine hydrate can be clearly observed. Finally, the XRD results confirm the formation of GO-based cubic spinel nickel ferrite nanocomposites: GONF and rGONF.

Fourier transfer infrared resonance (FT-IR) spectroscopy results (Fig. S1b†) of GO suggest the presence of epoxy (–C–O) and carbonyl (–C=O) functional groups with carboxylic groups. The peaks at 1054 and 1233 cm⁻¹ (–C–O), correspond to the stretching vibrations of epoxy and alkoxy groups. The sharp peaks at around 1420–1600 cm⁻¹ indicate the –C–C=C– stretching vibration of asymmetric aromatic alkenes in an aromatic ring. The sharp peaks at ∼1727 cm⁻¹ indicate the stretching vibrations of –C=O groups. The broad peaks at ∼3202 cm⁻¹ can be assigned to the hydrogen bond of the –OH group on aromatic phenol, alcohol, or carboxylic acids. The overall results of the FT-IR spectrum confirm the formation of GO from graphite.^41–44 The FT-IR spectra of GONF and rGONF indicate that the wavelengths of such functional groups in GO are shifted to lower wavelength regions upon magnetization of GO, or the reduction of GONF.⁴⁴ The shifting of the absorption peaks is associated with the formation of GONF and rGONF from GO. The sharp peaks at 550–500 cm⁻¹ indicate the tetrahedral and octahedral Fe–O and Ni–O bonds, and are clearly represented in Fig. S1b.†^43,44

The two prominent peaks at ∼1358 and ∼1606 cm⁻¹ in the Raman spectrum of GO (Fig. 2) correspond to the D and G band, respectively. Usually the G band corresponds to the E_2g mode associated with the vibrations of sp² carbon domains, while the D band is associated with the vibrations of sp³ carbon atoms in defects and disordered regions, that can break the symmetry and selection rules.⁴⁵ These strong D and G bands are very similar to previous results for GO characterization.⁴⁶ For the GONF nanocomposite, the D band shows a lower Raman shift (∼1310 cm⁻¹) with almost the same G band (∼1605.8 cm⁻¹) as GO, further confirming the formation of GONF by the interaction of NiFe₂O₄ with the sp² carbons (–C–O; –C=O; –C–C=C–) of GO. Meanwhile for rGONF, the D and G bands moved to a lower range Raman shift at ∼1303 and ∼1591 cm⁻¹ are associated with the reduction of GO. The formation of GONF and the reduction of rGONF were further confirmed by measuring their structural disorder from the intensity ratio of the D and G bands (I_D/I_G).⁴⁷ The intensity ratios of the D and G bands for GONF and rGONF are 1.08 and 1.094, respectively, showing an enhanced value compared to that of GO (0.96). This result indicates the presence of localized sp³ defects within the sp² carbon frame upon reduction or interaction of GO with NiFe₂O₄.³³

Fig. 2 Raman spectra of GO and GO based porous inverse spinel nickel ferrite nanocomposites: GONF and rGONF.

X-ray photoelectron spectroscopy (XPS) (Fig. 3a and b) studies were also performed to confirm the formation of GO, GONF and rGONF. The high binding energy intensity of the O (1s), Fe (2p) and Ni (2p) peaks (Fig. 3a) indicates the formation of GONF and rGONF from GO.⁴⁸ In addition, there is no indication of a satellite peak at ∼719 eV along with the peaks at around 711.00 eV (Fe 2p_3/2), 724.00 eV (Fe 2p_1/2), 855.00 eV (Ni 2p_3/2) and 861.00 eV (Ni 2p_1/2) (Fig. 3b). The XPS results indicate the presence of inverse spinel NiFe₂O₄ or the formation of GONF and rGONF from GO. The shifting of the O (1s) (Fig. 3a) binding energy to a lower range due to the loading of spinel NiFe₂O₄ onto GO can also be observed, indicating the formation of GONF and rGONF. The peaks at around 529.8 eV and 530.3 eV are due to oxygen in GONF and rGONF, respectively, while the peak at around 532.4 eV originates from oxygen in GO.⁹ The O (1s), Fe (2p) and Ni (2p) binding energies of rGONF are slightly more than those of GONF. This might be due to the decreased nuclear size or particle size of GONF caused by reduction.^43,44 The XPS results indicate the presence of cubic spinel nickel ferrite in GONF and rGONF.

Fig. 3 XPS full scan spectra of GO, GONF and rGONF (a) and XPS spectra of Fe (2p_3/2 and 2p_1/2), Ni (2p_3/2 and 2p_1/2) present in GONF and rGONF (b).

Scanning electron microscopy (SEM) images of GO, GONF and rGONF are shown in Fig. S2(a–d).† The obtained SEM surface images clearly indicate that porosity is increased as particle size decreases, along with a significant difference in surface morphology due to the formation of GONF and rGONF from GO. The obtained porous surface morphology is favorable for surface interface studies such as adsorption. Hence, the prepared GONF and rGONF can be used as adsorbents. Furthermore, the size, shape and morphology of the prepared adsorbents were confirmed using atomic force microscopy (AFM). A typical sample was prepared by dispersing the prepared sample in isopropyl alcohol and depositing it on a silica nitride AFM tip (Veeco Inc., NY). The AFM images of GONF and rGONF (Fig. 4a–d) reveal that the prepared GONF and rGONF composites are nano-sized with sphere shape particles. The formation of nano-sized particles upon the intercalation of nickel ferrite with GO leads to an increase in the surface area of the resultant magnetic spinel, GONF, in a wide range with an average particle size of 40.24 nm. The GONF particle size was reduced to 32.32 nm after the reduction of GONF.

Fig. 4 Atomic force microscopy (AFM) images of the nanocomposites: GONF and rGONF.

The surface characteristics of GONF and rGONF were studied using BET analysis with nitrogen (N₂) adsorption–desorption isotherms. The N₂ adsorption–desorption isotherms of GONF and rGONF (Fig. S3†) follow a typical type-IV isotherm with a characteristic hysteresis loop (H3 type), indicating the meso-porous nature of GONF and rGONF.^48,49 These hysteresis loops imply the conversion of GO to GONF and rGONF, which causes an increase in meso-porosity as the pore volume increases and the particle size decreases. The BET surface area of GO increases from 2.59 to 167.26 m² g⁻¹ as the particle size decreases from 1000 nm to 34.46 nm from the conversion of GO to rGONF (Table S2†). This may be due to the formation of the end product, rGONF, by the intercalation of cubic spinel NiFe₂O₄ with GO. The reduction or modification of GO with nickel ferrite increases its surface area to a large size and is favorable for enhancing its adsorption capacity. The significant changes in the surface texture of GO upon the intercalation of nickel ferrite lead to increases in the porosity with the increasing surface area of GONF and rGONF. The resultant average pore volume and pore diameter of GONF and rGONF indicate nano-crystalline meso-porous composite particles. The resultant average particle sizes, 44.12 nm for GONF and 34.46 nm for rGONF, are in close agreement with their average particle sizes obtained from AFM studies (40.24 nm for GONF and 32.32 nm for rGONF) and XRD Scherer crystallite sizes (41.0 nm for GONF and 32.0 nm for rGONF). From these overall results, the average particle size of the GONF and rGONF composites is ∼41.0 ± 2.0 and ∼32.0 ± 2.0 nm, respectively.

The magnetic hysteresis curves (M–H) of the nano-magnetic spinel GONF and rGONF composites were recorded at 300 and 5 K and are shown in Fig. 5a and b. The resultant magnetic properties, summarized in Table 1, indicate the superparamagnetic nature of rGONF and the ferromagnetic nature of GONF at room temperature. The superparamagnetic properties of rGONF were confirmed by the temperature dependent, M–T curves (Fig. 5d) obtained under a magnetic field of 1000 Oe. The zero field cooled (ZFC) and field cooled (FC) curves linearly increase with decreasing temperature when the blocking temperature (T_B) is 23 K, indicating the superparamagnetic properties of rGONF. Meanwhile, M–T curves (Fig. 5c) were not observed for GONF. This difference in the magnetic properties of GONF and rGONF may be explained by the smaller particle sizes of rGONF compared to GONF.^33,41,44 However, the high saturation magnetization (M_S) values of GONF (50.7 emu g⁻¹) and rGONF (31.6 emu g⁻¹) indicate that the prepared magnetic nanocomposites can be easily separated even with a small external magnetic field. The M_S value of the resultant nanocomposites, GONF and rGONF, was smaller than that of bulk NiFe₂O₄ (60 emu g⁻¹).⁵⁰ This result can be explained by the electron exchange between nickel ferrite and GO, causing the quenching of the magnetic momentum in GONF.^51,52 GONF or rGONF dispersed in water was magnetically separated in less than 10 s by an external magnetic field of ∼20 mT and the resultant photograph is shown in Fig. S4.†

Fig. 5 Magnetization vs. magnetic field (M–H) (a and b) and magnetization vs. temperature (M–T) (c and d) hysteresis curves for GONF and rGONF.

Table 1 Magnetic parameters measured by M–T hysteresis curves and M–T curves of GONF and rGONF

Nanocomposite	Temperature, K	MS, emu g^−1	HC, Oe	MR, emu g^−1	SR = MR/MS	χ, emu cm^−3 Oe^−1	TB, K
GONF	5	57.6	143	21.6	0.375	1.17 × 10^−3	240
	300	50.7	80	9.6	0.189	1.04 × 10^−3
rGONF	5	46.9	28	2.7	0.058	9.36 × 10^−4	23
	300	31.6	0.80	0.06	0.002	6.29 × 10^−4

---

3.2 Adsorption studies of As(III) and As(V) onto GONF and rGONF

The effect of contact time for the adsorption of As(III) and As(V) onto rGONF and GONF was investigated in order to evaluate the equilibrium point. The results are shown in Fig. 6a. From Fig. 6a, it can be clearly observed that the As(III) and As(V) adsorption was very fast during the first 10–15 min with 20–50% adsorption removal. The adsorption removal then slowly increased and reached equilibrium at 150 min. The maximum percentage removal of As(III) was 42% and 53% for GONF and rGONF, respectively. Meanwhile the maximum removal of As(V) was 79.2% and 99.7% for GONF and rGONF, respectively. The initial very fast adsorption of both As(III) and As(V) can be explained by the fact that GONF and rGONF have more freely available reactive sites with high surface area and relatively large pore sizes. The optimum equilibration time was determined to be 150 min for both arsenic ions.

Fig. 6 Contact time effect on the adsorption removal of As(III) and As(V) onto GONF and rGONF at 298 ± 2 K (a) and their adsorption kinetic models: pseudo-first order (PFO) (b), pseudo-second-order (PSO) (c), intra-particle diffusion model (d) and Boyd model (e). Adsorbent dosage = 0.3 g L⁻¹; initial metal ion concentration: 7.0 mg L⁻¹; pH: 6.5 (±0.20); equilibrium time: 150 min. To avoid any discrepancy or error, metal blank and three duplicate measurements were measured.

In order to understand the adsorption behavior of GONF and rGONF, adsorption kinetics data for both As(III) and As(V) (Fig. 6a) were evaluated by applying the linearized pseudo-first-order (PFO) and pseudo-second-order (PSO) kinetics models (for details see ESI†), as shown in Fig. 6b and c. The resultant kinetics parameters are summarized in Table 2. The regression coefficient (R²) and arsenic ion uptake capacity (q_e) (calculated and theoretical) (Table 2) indicate that the PSO kinetics is well fitted to the kinetic data for the adsorption of As(III) and As(V) onto GONF and rGONF. It was thus concluded that the adsorption kinetics of As(III) and As(V) onto GONF and rGONF is rate-limiting pseudo-second-order diffusion kinetics. However, the linear regression coefficients (R²) of the Webber–Morris intra-particle diffusion model (Fig. 6d) were determined to be 0.799, 0.829, 0.819 and 0.786 for the adsorption of As(III) and As(V) onto GONF and rGONF, which indicates that intra-particle diffusion is not main adsorption process. The regression plots of t^0.5 vs. q_t (Fig. 6d) are linear but do not pass through the origin. This result suggests that intra-particle diffusion was not the sole type of adsorption.^53,54 Further, the adsorption kinetic data were analyzed using the Boyd model.⁵⁵ The regression plot of t vs. B_t (Fig. 6e) is linear but does not pass through the origin. This indicates the occurrence of outer diffusion during the entire adsorption process.⁵⁶ However, some adsorption plots for the adsorption of As(III) and As(V) onto rGONF do pass through origin, suggesting the occurrence of both intra-particle diffusion and outer diffusion during the entire adsorption process. These results again confirm that intra-particle diffusion was not sole type of adsorption process.

Table 2 Kinetics parameters for the adsorption of both arsenic ions (7.0 mg L⁻¹) onto the GONF and rGONF (0.3 g L⁻¹) nanocomposites at pH 6.5 (±0.20) and 298 ± 2 K (N = 3, mean ± SD)

Kinetic parameter	GONF		rGONF
	As(III)	As(V)	As(III)	As(V)
qe,th (mg g^−1)	10.15 ± 0.12	20.50 ± 0.15	12.80 ± 0.18	22.48 ± 0.14
Pseudo-first-order, [log(qe − qt) = log qe − k1t]
k1 (min^−1)	0.027 ± 0.012	0.015 ± 0.011	0.015 ± 0.008	0.023 ± 0.007
qe,cal (mg g^−1)	7.72 ± 0.24	16.03 ± 0.52	9.64 ± 0.73	17.35 ± 0.46
R^2	0.835	0.977	0.970	0.911
Pseudo-second-order, [t/qt = 1/(k2qe^2) + t/qe]
k2 (g mg^−1 min^−1)	0.0043 ± 0.001	0.0019 ± 0.003	0.0033 ± 0.005	0.0028 ± 0.002
qe,cal (mg g^−1)	10.76 ± 0.05	21.88 ± 0.01	13.58 ± 0.02	23.47 ± 0.02
R^2	0.999	0.997	0.997	0.999

---

In order to estimate the adsorption capacity and adsorption nature of the nanocomposites, three isotherm models: Langmuir, Freundlich and Temkin, were used to fit the adsorption isotherm data for As(III) and As(V) adsorbed onto rGONF and GONF (Fig. 7a) at equilibrium. The applied linear Langmuir (eqn (1)), Freundlich (eqn (2)) and Temkin (eqn (3)) isotherm models are shown here:

C_e/q_e = 1/q_mK_L + C_e/q_m (1)

log q_e = log K_F + n⁻¹ log C_e (2)

q_e = B log K_T + B log C_e (3)

where q_m and K_L, n, K_F and B, and B log K_T were calculated from the slope and intercept of the linear curves of C_e/q_e vs. C_e (eqn (1)), log C_e vs. log q_e (eqn (2)) and q_e vs. log C_e (eqn (3)), respectively. The obtained parameters from the linear fits of the three isotherm models for As(III) and As(V) adsorption are summarized in Table 3. The resultant correlation coefficient (R²) values indicate that the Langmuir isotherm was well tuned to the adsorption data for both As(III) and As(V) on rGONF and GONF compared to the other two isotherm models. Moreover, the Freundlich constant, n, being more than 1, indicates the favorable adsorption of As(III) and As(V) on the prepared nanocomposites. The Langmuir separation constant (R_L < 1.0) indicates the favorability of the Langmuir isotherm for the As(III) and As(V) adsorption onto GONF and rGONF. From Table 3, the maximum adsorption capacities of As(III) and As(V) for rGONF are much higher than those of GONF (59.52 and 81.30 mg g⁻¹ for As(III) and As(V), respectively), rGO (43.78 and 52.18 mg g⁻¹ for As(III) and As(V), respectively), GO (28.25 and 42.02 mg g⁻¹ for As(III) and As(V), respectively), and NiFe₂O₄ (18.58 and 32.78 mg g⁻¹ for As(III) and As(V), respectively). It can be interpreted that rGONF has a high surface area with relatively large pore sizes (Table S1†). The isotherm results indicate that the adsorption of both As(III) and As(V) is a monolayer adsorption that occurs at binding sites on the homogeneous surface of the adsorbent. Furthermore, compared with other adsorbents shown in Table 4, the adsorption capacities of GONF and rGONF are higher than or comparable to those reported in the literature. The difference in the adsorption performance of As(III) and As(V) might be due to differences in the pK_a values in aqueous solutions. Due to the pK_a values, As(V) can be negative under all pH conditions, which causes favorable adsorption onto the positive surfaces of nanocomposites at around pH 6.5. The maximum adsorption capacities of GONF and rGONF for As(III) and As(V) at different temperatures (283 to 323 ± 2 K) are shown in Fig. 7b. The resultant maximum adsorption capacities of rGONF and GONF for both As(III) and As(V) are increased with increasing temperature from 283 to 300 ± 2 K, while upon a further increase in the temperature, the adsorption removal capacities of both As(III) and As(V) are decreased. A similar observation has been reported in literature, i.e., an increase in the adsorption removal of arsenic from 283 to 300 K,^57,58 but a decrease in the adsorption removal as the temperature is increased above 300 K.^59,60 From the obtained results, the optimum temperature required for the maximum adsorption of arsenic ions is 300 K.

Fig. 7 Adsorption isotherms of As(III) and As(V) onto the nanocomposites, GONF and rGONF, at 293 ± 2 K (a), and the temperature effects on the adsorption capacities of the nanocomposites with arsenic ions at 283 to 323 ± 2 K (b). pH: 6.5 (±0.2); initial As(III) & As(V) concentrations: 7.0 mg L⁻¹; time: 150 min; GONF and rGONF dosage: 0.3 g L⁻¹.

Table 3 Adsorption isotherm parameters for As(III) and As(V) adsorption onto GONF and rGONF at pH 6.5 (±0.20) and 298 ± 2 K for 150 min (N = 3, mean ± SD)

Isotherm models	GONF		rGONF
	As(III)	As(V)	As(III)	As(V)
Langmuir
qm (mg g^−1)	59.52 ± 0.50	81.30 ± 0.40	65.78 ± 0.31	106.40 ± 0.30
KL (L mg^−1)	0.30 ± 0.02	0.57 ± 0.01	0.53 ± 0.03	0.35 ± 0.02
R^2	0.990	0.999	0.998	0.999
RL	0.47 ± 0.02	0.25 ± 0.01	0.27 ± 0.006	0.28 ± 0.005
Freundlich
KF (mg g^−1 (L mg^−1)^1/n)	14.43 ± 0.63	25.92 ± 0.72	20.64 ± 0.52	67.79 ± 0.55
N	1.99 ± 0.05	1.87 ± 0.02	2.04 ± 0.04	2.16 ± 0.03
R^2	0.966	0.944	0.921	0.910
Tempkin
KT, L mg^−1	3.57 ± 0.06	6.43 ± 0.04	19.19 ± 0.05	44.63 ± 0.05
B (g L^−1)	28.25 ± 0.62	39.35 ± 0.56	31.67 ± 0.52	44.58 ± 0.58
R^2	0.986	0.995	0.987	0.975

---

Table 4 Comparison of the adsorption capacities of GONF and rGONF with other literature reported adsorbents for As(V) and As(III)

Name of adsorbent	Metal ion removal, mg g^−1
	As(V)	As(III)
Magnetite nanoparticles^7	3.7	3.7
Fe3O4–RGO–MnO2 nanoparticles^8	12.22	14.04
Fe3O4–RGO nanoparticles^9	10.2	13.1
Biochar/γ-Fe2O3 composite^11	3.1	—
NZVI nanoparticles^12	—	3.5
Ascorbic acid-coated-Fe3O4 nanoparticles^21	16	46
Natural iron ores^22	0.4	—
NZVI–activated carbon nanoparticles^23	12.0	18.2
NZVI–RGO nanoparticles^66	29.04	35.83
Magnetic wheat straw (MWS)^67	8.02	3.8
Fe3O4–HBC composite^68	3.35	3.07
Iron-containing ordered mesoporous carbon^66	7.0	9.3
rGONF (this work)	106.40	65.78
GONF (this work)	81.30	59.52

---

A solution pH ranging from 2.0 to 9.0 was used for investigating the effect of pH on the adsorption potential of GONF and rGONF for As(III) and As(V), as shown in Fig. 8a. The adsorption efficiency of As(III) increased sharply from pH 2.0 to 6.0 and then reached maximum adsorption between pH 6.0 and 8.0. Meanwhile, As(V) shows a maximum adsorption at pH 2.0 and then slowly decreases to a lower range with increasing the pH from 3.0 to 9.0. Similar tends have been observed in previous reports.^9,58 This adsorption phenomenon can be explained by the species of arsenic ion and the pH_zpc of GONF (6.4) and rGONF (6.2) (the pH_zpc are measured using a well-known acid-base titration method at 0.01 mol L⁻¹ ionic strength and room temperature). As(III) presents in a non-ionic form at pH < 9.1 and in a negative ionic form above pH 9.1, while As(V) presents in a negative ionic form under most pH conditions.⁶¹ At pH < pH_zpc, the GONF and rGONF surfaces are positively charged, while at a pH > pH_zpc, the GONF and rGONF surfaces are negatively charged. Under acidic conditions, most of the surface of the nanocomposites is positively charged, while As(III) presents as a mixture of cationic and a neutral species. As the surface of the nanocomposites and As(III) present as mostly neutral species between pH 6 and 8, there is little repulsion between adsorbent and adsorbate, causing the favorable adsorption of As(III) onto the nanocomposites. On the other hand, the fraction of negatively charged arsenic species and negative surface of the nanocomposites increases at higher pH, causing decreased adsorption of As(III) onto the nanocomposites. The increase in the adsorption of As(III) with increasing pH to alkaline conditions and the higher adsorption of As(III) at pH > pH_zpc suggests that the adsorption occurs through both a surface-complexation reaction and electrostatic attraction. Similar observations have been found in previous studies.^58,62

Fig. 8 Studies on the effects of aqueous solution pH (a) and dosage (b) on the adsorption removal of As(III) and As(V) with GONF and rGONF. Initial metal ion concentration: 7.0 mg L⁻¹; equilibrium time: 150 min; adsorbent dosage = 0.3 g L⁻¹ (for pH effect study) temperature: 298 ± 2 K.

For As(V), the decreased adsorption as the pH increases is due to the net positive charge of the nanocomposite at pH < pH_zpc, which attracts negatively charged As(V) at lower pH values resulting greater adsorption, whereas by increasing the pH, the decreased positive charge on the surface sites at pH > pH_zpc causes a decrease in the percentage of As(V) adsorbed due to electrostatic repulsion. However, because significant As(V) adsorption was observed above pH_zpc, where both adsorbate and adsorbent are present as negative charges, the surface-complexation reaction may be the main factor involved in the adsorption of As(V) onto the nanocomposites, along with electrostatic repulsion. Similar results have been observed in previous reports.^9,58,62

As shown in Fig. 8b, the removal percentage of As(III) (26.3–95.5% with rGONF and 15–84.7% with GONF) was rapidly increased as the dosage of nanocomposites increased from 0.1 to 0.6 g L⁻¹. Then, it was slowly increased (95.5–99.6% with rGONF and 84.7–99.2% with GONF) from 0.6 to 1.0 g L⁻¹ and reached a maximum (99.6% with rGONF and 99.2% with GONF) at 1.0 to 2.0 g L⁻¹. Meanwhile for As(V), the removal percentage was rapidly increased (53.2–99.6% for rGONF) as the rGONF dosage was increased from 0.1 to 0.3 and then reached a maximum (99.6%) at 0.3 g L⁻¹ of rGONF. However, the removal percentage of As(V) by GONF slowly increased as the dosage was increased from 0.1 g L⁻¹ to 0.5 g L⁻¹ and reached a maximum adsorption (99.99%) at 0.5 g L⁻¹. Above 1.0 g L⁻¹, the adsorption removal of arsenic ions was almost constant. The difference in the adsorption capacity of both nanocomposites for As(III) and As(V) may be explained by the different sizes and surface areas of the nanocomposites, along with the nature of the arsenic ion species. The effect of the ionic strength or background ion effect on the adsorption of As(III) and As(V) onto GONF and rGONF was studied using NaCl, KCl, NaNO₃, CaCl₂ and MgCl₂ as background electrolytes, ranging from 0.005 to 0.5 mol L⁻¹ and the results are shown in Fig. S5.† The experimental results indicate that no significant change was observed in the adsorption of As(III) and As(V) onto GONF or rGONF. From Fig. S5,† the concentration of background cations such as Na⁺, K⁺, Ca²⁺ and Mg²⁺ does not show a significant effect on the adsorption process for both arsenic ions. The effect of other co-existing anions, such as SO₄²⁻, Cl⁻, F⁻, PO₄³⁻, NO₃²⁻ and CO₃²⁻ (0.05 mol L⁻¹), on the arsenic (10 mg L⁻¹) adsorption was studied under the same experimental conditions and the results are shown in Fig. 9. It was observed that the co-existing anions other than phosphate did not show a significant effect on the adsorption removal of both arsenic ions. The small effect of Cl⁻, F⁻, NO₃⁻ and CO₃²⁻ on the arsenic adsorption may be attributed to an inner-sphere complexation between arsenic and the nanocomposites.⁶³ The slight decreases in the As(III) adsorption in the presence of SO₄²⁻ may be due to the competitive adsorption for complexation at the surface sites of the adsorbents.⁵³ The phosphate ion caused the greatest decrease in the adsorption removal by GONF and rGONF for both arsenic ions. This might be explained by the fact that arsenate and phosphate are tetrahedral anions that compete with each other for active adsorption sites on the GONF (surface sites or charge density: 9.81 × 10⁻⁴ mol g⁻¹) and rGONF (surface sites or charge density: 7.14 × 10⁻⁴ mol g⁻¹) surfaces, resulting in decreased adsorption capacity.^53,64 Another previous study reported that As(III) and As(V) can form inner sphere complexes as well as outer-sphere ones with iron oxide coated nanocomposites.^53,63

Fig. 9 Co-existing ion effect on adsorption removal of As(III) (a) and As(V) (b) with GONF and rGONF (0.3 g L⁻¹). As(III) and As(V) concentration: 10 mg L⁻¹; co-existing ions concentration: 0.05 mol L⁻¹; pH: 6.5; equilibrium time: 150 min; temperature: 298 ± 2 K.

3.3 Adsorption mechanism

Based on the XPS spectrum in Fig. 10, showing the nanocomposites before and after adsorption of As, a shifting of the O (1s), C (1s), and Fe and Ni (2p) peaks is clearly observed. These results indicate the adsorption of As by complexation or electrostatic interactions with nickel ferrite and C=O, C–O and O–C=O functional groups on the nanocomposite surfaces at around pH 6.5. This adsorption phenomenon can be explained by the species of arsenic ion and the pH_zpc of GONF (6.4) and rGONF (6.2) (the pH_zpc are measured by a well-known acid–base titration method at 0.01 M ionic strength and room temperature). At pH < pH_zpc, the GONF and rGONF surfaces are positively charged, while at a pH > pH_zpc, the GONF and rGONF surfaces are negatively charged. As(III) presents in a non-ionic form at pH < 9.1 and in a negative ionic form above pH 9.1, while As(V) presents in a negative ionic form under most pH conditions.⁶¹ Hence, this negative As(V) and neutral As(III) species at pH 6.5 might adsorb onto the positive surface sites of the nanocomposites, GONF (surface sites density: 9.81 × 10⁻⁴ mol g⁻¹) and rGONF (surface sites density: 7.14 × 10⁻⁴ mol g⁻¹), by complexation or electrostatic interactions. The possible complexation or electrostatic interaction mechanisms between As and the nanocomposites under pH ≤ pH_zpc in aqueous solution are:

S-OH + H₃O⁺ → SOH₂⁺ + H₂O (under pH ≤ pH_zpc) (4)

S-OH + OH⁻ → SO⁻ + H₂O (under pH ≥ pH_zpc) (5)

SOH₂⁺ + H₂AsO₄⁻ → SOH₂⁺ − H₂AsO₄⁻ (6)

SOH₂⁺ + H₂AsO₃⁻ → SOH₂⁺ − H₂AsO₃⁻ (7)

Fig. 10 Comparison of XPS spectra before and after arsenic loading of the nanocomposites to evaluate the adsorption mechanism. Full XPS scan of GONF, rGONF alone and arsenic loaded (a), high resolution O (1s) XPS scan of GONF, rGONF alone and arsenic loaded (b), high resolution C (1s) of GONF and GONF-As (c) and high resolution C (1s) rGONF and rGONF-As (d).

3.4 Regeneration and re-usability of the nanocomposites: GONF and rGONF

The stability and reusability of the prepared nanocomposites were investigated by the adsorption and desorption of arsenic ions onto GONF and rGONF. 10 mg L⁻¹ of both As(III) and As(V) loaded onto GONF and rGONF were recovered by an external magnetic field and treated with 0.2 M NaOH aqueous solution at each cycle. The adsorption removal percentage of the regenerated GONF and rGONF are shown in Fig. 11. It can be seen that the adsorption capacity of GONF and rGONF for both arsenic ions was stable for up to five cycles but decreased after the 6th cycle, indicating that GONF and rGONF can be reused up to five cycles repeatedly. It means that the surface of GONF and rGONF may be becoming destroyed due to some of the nickel ferrite particles dissolving after the fifth cycle. The desorption capacities of GONF and rGONF were also good for up to five cycles, at more than 95%, but this became less than 80% after the 6th cycle. However, the desorption of As(III) and As(V) from the loaded nanocomposites was tested at different concentrations of NaOH (0.1, 0.2, 0.3, 0.4, 0.5 and 0.6 M) for the complete desorption of arsenic species. From the results, an increase was observed in the desorption percentage (from 90 to 99.98%) of arsenic species, when the concentration of the desorbing agent was increased from 0.1 to 0.4 M. Thereafter, the desorption percentage was constant, as the desorbing agent concentration was increased from 0.4 to 0.6 M. However, if we use 0.4 M NaOH as the desorbing agent for re-cycling the nanocomposites, the re-usability capacity for the adsorption of arsenic decreases and the nanocomposites can be re-used only up to two times (with 99.98% adsorption). This may be explained by the desorption of nickel ferrite, or the increased negative charge on the surface of the nanocomposites by strong binding with high concentration NaOH, which may cause a decrease in the adsorption removal from 75% to 50% over three re-cycles. Moreover, a 95% desorption of arsenic with almost constant re-adsorption (around 99.8%) was achieved with 0.2 M NaOH for up to 5-cycles of nanocomposite re-use. Hence, we selected 0.2 M NaOH as a desorbing agent.

Fig. 11 Regeneration and re-usability test results of the nanocomposites: GONF and rGONF (‘*’ and ‘**’ indicate the slight and significant changes in the adsorption removal of arsenic species, respectively).

3.5 Solubility of Ni and toxicity test of the nanocomposites

In order to examine the solubility of Ni–ferrite in the nanocomposites, 1.0 g of each nanocomposite (GONF and rGONF) was added to 1 L double distilled water. The experimental results found 0.064 mg L⁻¹ and 0.012 mg L⁻¹ of Ni leaking from the GONF and rGONF, respectively. This solubility is less than the EPA permissible level for Ni (0.1 mg L⁻¹) in drinking water.⁶⁵

3.6 Application to real groundwater spiked with arsenic

In order to test the applicability of GONF and rGONF for the adsorption removal of arsenic from real groundwater (collected from Hagye-dong, Nowon-gu, Seoul, RO Korea), different concentrations of As (2.0, 4.0, and 8.0 mg L⁻¹) were spiked and the removal results are shown in Table 5. Both As ions were almost completely adsorbed from the spiked real samples. After adsorption removal of the As ions, the remaining As concentrations in the real samples were below the EPA permissible levels (0.01 mg L⁻¹ to 0.05 mg L⁻¹).⁶ This result suggests that the nanocomposite used in this work can be successfully applied for the treatment of real groundwater containing arsenic.

Table 5 Adsorption removal of As(III) and As(V) from As spiked real groundwater, collected from Hagye-dong, Nowon-gu, Seoul, Korea (N = 3, mean values are reported here, the standard deviation is less than 5%)

Name of sample	As(III) or As(V) added, mg L^−1	Adsorption of As(III), mg g^−1		Adsorption of As(V), mg g^−1
		GONF	rGONF	GONF	rGONF
Sample 1	2.0	1.953	1.962	1.947	1.961
Sample 2	4.0	3.945	3.946	3.972	3.955
Sample 3	8.0	7.953	7.945	7.950	7.981

---

4. Conclusions

The mesoporous nano-crystalline cubic inverse spinel structure of the present magnetic nanocomposites, GONF and rGONF, was confirmed by the XRD, FT-IR, and XPS characterization results. SEM, AFM and BET analyses verified the formation of GONF and rGONF with a porous surface morphology, having BET surface areas of 123 m² g⁻¹ and 167 m² g⁻¹, respectively. XRD, BET and AFM studies concluded that the average sizes of the GONF and rGONF particles were 41 nm and 32 nm, respectively. The resultant magnetic measurements confirmed the ferromagnetic nature of GONF and the superparamagnetic nature of rGONF. The adsorption characteristics confirmed that the present nanocomposites are suitable for the adsorption removal of As(III) and As(V), with considerable enhanced adsorption capacity compared to that of GO and other reported nanocomposite materials. It also verified that the adsorption of rGONF is higher than that of GONF for both arsenic ions. The adsorption kinetics and isotherm results confirmed that the adsorption of As(III) and As(V) was a monolayer adsorption on the homogenous surface of the nanocomposites, GONF and rGONF, with following pseudo-second-order adsorption kinetics. However, the Boyd and intra-particle diffusion model confirmed that the adsorption comprised both inner-sphere and outer-sphere complexation. The adsorption of As(III) and As(V) increases with increasing temperature at a certain temperature (300 K), while on further increasing the temperature, the adsorption removal of both arsenic ions decreases. The pH and ionic strength results indicated that the adsorption of As(III) and As(V) onto GONF and rGONF occurred through both surface complexation and electrostatic interaction. The stability and regeneration studies confirmed that the present nanocomposites were well suited for the removal of As(III) and As(V) for up to 5 cycles without loss of capacity. The nanocomposites were successfully applied for arsenic removal from real groundwater spiked with arsenic. Both As ions were almost completely removed from the spiked real samples. After adsorption removal of the As ions, the remaining As concentration was below the EPA permissible levels.

Author contribution

JKY and JRK conceived and wrote the manuscript, designed the experiments and analyzed the data. LPL, YLC and ISK performed the research. YYC supervised the study/project.

Conflicts of interest

Competing financial interests: the authors declare no competing financial interests.

Acknowledgements

This work was supported by the Korea Ministry of Environment as the “GAIA project (2014000550003)” and was partly supported by a Research grant-2016, Kwangwoon University, Seoul, Korea.

References

1. D. Mohan and C. U. Pittman Jr, J. Hazard. Mater., 2007, 142, 1–53.
2. L. Lorenzen, J. S. J. Van Deventer and M. W. Landi, Miner. Eng., 1995, 8(4), 557–569.
3. H. Erdogan, O. Yalcınkaya and A. R. Turker, Desalination, 2011, 280, 391–396.
4. S. Chakravarty, V. Dureja, G. Bhattacharyya, S. Maity and S. Bhattacharjee, Water Res., 2002, 36, 625–632.
5. Concept note on Geogenic Contamination of Ground water in India, Central Ground Water Board, Ministry of water Resources, Govt. of India, 2014.
6. WHO, Guidelines for drinking water quality Recommendations, World Health Organization, Geneva, 2nd edn, 1993, vol. 1.
7. A. I. Zouboulis and I. A. Katsoyiannis, Environ. Int., 2005, 31(2), 213–219.
8. X. B. Luo, C. C. Wang and S. L. Luo, Chem. Eng. J., 2012, 18745–18752.
9. V. Chandra, J. Park, Y. Chun, J. W. Lee, I. C. Hwang and K. S. Kim, ACS Nano, 2010, 4, 3979–3986.
10. A. Navas-Acien, A. R. Sharrett, E. K. Silbergeld, B. S. Schwartz, K. E. Nachman, T. A. Burke and E. Guallar, Am. J. Epidemiol., 2005, 162, 1037–1049.
11. M. Zhang, B. Gao, S. Varnoosfaderon, A. Hebard, Y. Yao and M. Inyang, Bioresour. Technol., 2012, 130, 457–462.
12. S. R. Kanel, B. Manning, L. Charlet and H. Choi, Environ. Sci. Technol., 2005, 39(5), 1291–1298.
13. H. Dong, X. Guan and I. Lo, Water Res., 2012, 46, 4071–4080.
14. R. Sandoval, A. M. Cooper, K. Aymar, A. Jain and K. Hristovski, J. Hazard. Mater., 2011, 193, 296–303.
15. P. Pillewan, S. Mukherjee, T. Roychowdhury, S. Das, A. Bansiwal and S. Rayalu, J. Hazard. Mater., 2011, 186, 367–375.
16. R. M. Dhoble, S. Lunge, A. Bhole and S. Rayalu, Water Res., 2011, 45, 4769–4781.
17. A. Sperlich, A. Werner, A. Genz, G. Amy, E. Worch and M. Jekel, Water Res., 2005, 39(6), 1190–1198.
18. K. P. Raven, A. Jain and R. H. Loeppert, Environ. Sci. Technol., 1998, 32, 344–349.
19. W. Driehaus, M. Jekel and U. Hildebrandt, Aqua, 1998, 47, 30–35.
20. S. Sarkar, L. M. Blaney, A. Gupta, D. Ghosh and A. K. Sen Gupta, Environ. Sci. Technol., 2008, 42, 4268–4273.
21. L. Feng, M. Cao, X. Ma, Y. Zhu and C. Hu, J. Hazard. Mater., 2012, 217, 439–446.
22. W. Zhang, P. Singh, E. Paling and S. Delides, Miner. Eng., 2004, 17, 517–524.
23. H. Zhu, Y. Jia, X. Wu and H. Wang, J. Hazard. Mater., 2009, 172(2–3), 1591–1596.
24. J. P. Jolivet, C. Chaneac and E. Tronc, Chem. Commun., 2004, 5, 481–487.
25. S. Zhang, X. Li and J. P. Chen, Carbon, 2010, 48, 60–67.
26. M. S. Mauter and M. Elimelech, Environ. Sci. Technol., 2008, 42, 5843–5859.
27. G. M. Scheuermann, L. Rumi, P. Steurer, W. Bannwarth and R. Malhaupt, J. Am. Chem. Soc., 2009, 131, 8262–8270.
28. Y. J. Yu, Y. Zhao, S. Ryu, L. Brus, K. S. Kim and P. Kim, Nano Lett., 2009, 9, 3430–3434.
29. S. Myung, J. Park, H. Lee, K. S. Kim and S. Hong, Adv. Mater., 2010, 22, 2045–2049.
30. H. P. Cong, J. J. He, Y. Lu and S. H. Yu, Small, 2009, 6, 169–171.
31. F. Yu, S. Sun, J. Ma and S. Han, Phys. Chem. Chem. Phys., 2015, 17, 4388–4397.
32. Y. Yao, S. Miao, S. Liu, L. P. Ma, H. Sun and S. Wang, Chem. Eng. J., 2012, 184, 326–332.
33. S. Bai, X. Shen, X. Zhong, Y. Liu, G. Zhu, X. Xu and K. Chen, Carbon, 2012, 50(6), 2337–2346.
34. X. Z. Sun and D. X. Li, Adv. Mater. Res., 2013, 834–836, 458–461.
35. R. J. Su, M. C. Zhu, X. Z. Sun and J. Guan, Appl. Mech. Mater., 2015, 768, 3–9.
36. B. Li, H. Cao, G. Yin, Y. Lu and J. Yin, J. Mater. Chem., 2011, 21, 10645–10648.
37. T. S. Sreeprasad, S. M. Maliyekkal, K. P. Lisha and T. Pradeep, J. Hazard. Mater., 2011, 186(1), 921–931.
38. L. P. Lingamdinne, J. R. Koduru, Y.-L. Choi, Y.-Y. Chang and J.-K. Yang, Hydrometallurgy, 2015 DOI:10.1016/j.hydromet.2015.11.005.
39. K. Maaz, S. Karim, A. Mumtaz, S. K. Hasanain, J. Liu and J. L. Duan, J. Magn. Magn. Mater., 2009, 321, 1838–1842.
40. S. W. Hummers Jr and R. E. Offeman, J. Am. Chem. Soc., 1958, 80(6), 1339.
41. Z. J. Fan, K. Wang, T. Wei, J. Yan, L. P. Song and B. Shao, Carbon, 2010, 48(5), 1670–1692.
42. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, ACS Nano, 2010, 4(8), 4806–4814.
43. B. Senthilkumar, R. Kalai Selvan, P. Vinothbabu, I. Perelshtein and A. Gedanken, Mater. Chem. Phys., 2011, 130, 285–292.
44. K. Nejati and R. Zabihi, Chem. Cent. J., 2012, 6(23) DOI:10.1186/1752-153x-6-23.
45. A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth and A. K. Geim, Phys. Rev. Lett., 2006, 97(18), 187401.
46. G. Zhao, J. Li, X. Ren, C. Chen and X. Wang, Environ. Sci. Technol., 2011, 45, 10454–10462.
47. C. N. R. Rao, A. K. Sood, K. S. Subramanyam and A. Govidaraj, Angew. Chem., Int. Ed., 2009, 48(42), 7752–7777.
48. J. Lu, X. Jiao, D. Chen and W. Li, J. Phys. Chem. C, 2009, 113, 4012–4017.
49. S. Purwajanti, L. Zhou, Y. Ahmad Nor, J. Zhang, H. Zhang, X. Huang and C. Yu, ACS Appl. Mater. Interfaces, 2015, 7(38), 21278–21286.
50. D. H. Chen and S. H. Wu, Chem. Mater., 2000, 12, 1354–1360.
51. M. H. Liao and D. H. Chen, J. Mater. Chem., 2002, 12, 3654–3659.
52. D. A. Vanleeuwen, J. M. Vanruitenbeek, L. J. Dejongh, A. Ceriotti, G. Pacchioni, O. D. Haberlen and N. Rosch, Phys. Rev. Lett., 1994, 73, 1432–1435.
53. F. Yu, S. Sun, J. Ma and S. Han, Phys. Chem. Chem. Phys., 2015, 17, 4388–4397.
54. M. Dogan, H. Abak and M. Alkan, J. Hazard. Mater., 2009, 164, 172–181.
55. M. S. Ray, Adsorpt. Sci. Technol., 2000, 18, 439–468.
56. V. Gupta, A. Mittal, V. Gajbe and J. Mittal, J. Colloid Interface Sci., 2008, 319, 30–39.
57. D. Pokhrel and T. Viraraghavan, J. Hazard. Mater., 2008, 150(3), 818–825.
58. C. Wang, H. Luo, Z. Zhang, Y. Wu, J. Zhang and S. Chen, J. Hazard. Mater., 2014, 268, 124–131.
59. P. Mondal, C. Balomajumder and B. Mohanty, J. Hazard. Mater., 2007, 144, 420–426.
60. F. Partey, D. Norman, S. Ndur and R. Nartey, J. Colloid Interface Sci., 2008, 321(2), 493–500.
61. T. F. Lin and J. K. Wu, Water Res., 2001, 35, 2049–2057.
62. X. Guo and F. Chen, Environ. Sci. Technol., 2005, 39, 6808–6818.
63. J. Wang, W. Xu, L. Chen, X. Huang and J. Liu, Chem. Eng. J., 2014, 251, 25–34.
64. A. Sigdel, J. W. Park, H. Kwak and P. K. Park, J. Ind. Eng. Chem., 2016, 35, 277–286.
65. WHO, Nickel, in Drinking-water: guidelines for drinking water quality, World Health Organization, Geneva, 2005.
66. Z. M. Gu, B. L. Deng and J. Yang, Materials, 2007, 102(1–3), 265–273.
67. Y. Tian, M. Wu, X. Lin, P. Huang and Y. Huang, J. Hazard. Mater., 2011, 193, 10–16.
68. S. A. Baig, T. T. Sheng, C. Sun, X. Q. Xue, L. S. Tan and X. H. Xu, PLoS One, 2014, 9(6), e100704.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra10134h

---