Magnetic modified graphene oxide was synthesized as a new modified magnetic nano-composite (MMNC) by a simple sonochemical–hydrothermal method. The sonochemical reaction was employed to exfoliate, functionalize and decorate neomycin on graphene oxide sheets. Nickel ferromagnetic particles were synthesized by hydrothermal co-precipitation method and decorated on neomycin-modified graphene oxide. The morphology and chemical structure of MMNC were characterized by scanning electron microscopy, energy dispersive spectroscopy and X-ray diffraction spectroscopy. The adsorption capability of MMNC for removal of phenolic compounds was assessed through adsorption of 4-nitrophenol (4-NP) from aqueous solution. The three-factor Box–Behnken design coupled with response surface method was applied to evaluate and optimize the important variables which affect the adsorption process. A significant quadratic model (p-value <0.05, R2(adj) = 0.9593) was derived using analysis of variance. The maximum adsorption capacity of 125.4 mg 4-NP/g MMNC at pH 6 was obtained, which was comparable in some cases and higher than most adsorbents reported in the literature. The presence of neomycin on graphene oxide sheets improved the maximum adsorption capacity of the nano-sorbent up to 28% (from 98.7 to 125.4 mg 4-NP/g adsorbent). The adsorption isotherms fitted well with the Langmuir model (Langmuir constant b = 0.064 l/mg, R2 = 0.9989) and the kinetic study showed that the nitrophenol uptake process followed the pseudo-second-order rate expression (R2 ≥ 0.9960, pseudo-second-order constant K2 ≥ 1.7 × 10−3).
Nitrophenol is one of the phenolic derivatives that extensively is used in the organic synthesis and production of pharmaceuticals, polymers, fungicides, pesticides and dyes and also to darken leather. Phenol and phenolic compounds are known for their hazardous effects on the environment and humans (Khaloo et al. 2015). The US Environmental Protection Agency (US EPA) has listed nitrophenol (NP) compounds as the most toxic pollutants due to their excellent solubility in water and high toxicity (US EPA 2017). Most phenolic compounds such as chlorophenols, aminophenols, chlorocatechols, NPs, methylphenols and other phenolic compounds have all been characterized as exerting toxic influence on humans. These compounds can easily penetrate the skin through absorption and can readily be absorbed from the gastrointestinal tract of humans. Once in the system, they undergo metabolism and transform to various reactive intermediate forms, particularly quinone moieties, which can easily form covalent bonds with proteins, resulting in their ability to exert toxic effects on humans (William et al. 2017). Due to the wide application of NP compounds and hence their widespread production, disposal of these materials in aquatic systems causes severe environmental problems.
Various technologies have been used for the removal of organic or inorganic pollutant from wastewater and groundwater including electrochemical removal (Pedersen et al. 2018), advanced oxidation process (Sivagami et al. 2018), coagulation (Abtahi et al. 2017; Gholami et al. 2017), ion exchange (López-Ortiz et al. 2018), precipitation (Silva et al. 2017), photocatalysis (Sharma et al. 2019) and adsorption (Yang et al. 2018). Among them, the adsorption technique has attracted huge attention in water treatment due to high removal efficiency without yielding harmful by-product (Naddafi et al. 2016; Nasseri et al. 2018). In the last decade, some novel adsorbents have been introduced for the adsorptive removal of organic and inorganic pollutants from aqueous environment, such as metal organic frameworks (Saedi & Roushani 2018), biomass (Altowayti et al. 2019), carbon-based nanomaterials (Khaloo et al. 2016) and alumina (Pham et al. 2018; Chu et al. 2019; Pham et al. 2019). Recently magnetic nano-adsorbent has gained increasing research in adsorptive water treatment techniques since the separation process could be done by applying an external magnet without filtration or centrifugation. Rapid mass transfer can be obtained due to the nano-sized structure of adsorbent having a large surface area for adsorption of analytes and, hence, facilitating the rapid adsorption equilibrium (Khaloo & Fattahi 2014).
Graphene, a two-dimensional carbon nanostructure, has attracted numerous investigations due to its unique chemical, physical and mechanical properties and a wide range of graphene-based nanomaterials have been studied and applied in various fields (Kaur et al. 2018). The combination of graphene and magnetic nano-particles offers magnetic graphene nano-composites as new material and provides new ways to develop novel catalysts, electrode materials and adsorbents (Hu et al. 2017; Liu et al. 2017; Rashidi Nodeh et al. 2017; Yang et al. 2017).
Recently, considerable attention has been directed towards the metal ferrite spinel compounds of the type MFe2O4 (M = Mn, Ni, Co, Cu, Zn). Ferrite spinel compounds are well-known magnetic materials with high magnetic permeability, large expansion coefficient, low magnetic transition temperature, high adsorption capacities, and low magnetic losses and with a wide potential application in water and wastewater treatment (Kefeni et al. 2017). With a combination of graphene oxide nanosheets and ferrite spinel compounds, new composites of nanomaterials have been launched which have the benefits of these two groups together. Therefore in recent years, graphene-based magnetic compounds have been shown to have a wide application in remediation of the environment (Li et al. 2018; Sherlala et al. 2018).
Different methods could be used to study the effect of various chemico-physical parameters on the efficiency of a process in chemistry. Central composite design (CCD) coupled with response surface method (RSM) is one of the useful methods for optimizing the important variables affecting a process. RSM has been applied to many purification processes such as coagulation process (Adesina et al. 2019), electrochemical process (Sharma & Simsek 2020), adsorptive removal (Gadekar & Ahammed 2019) and photo-Fenton oxidation (Abd Manan et al. 2019). CCD is a very efficient technique, providing much information on experiment variable effects and overall experimental error in a minimum number of required runs, while RSM is a collection of statistical and mathematical techniques based on the fit of the experimental data in a polynomial equation, which must describe the behavior of a data set with statistical model previsions. Therefore, the combination of CCD and RSM creates a statistical technique for designing experiments, building models, evaluating the effects of several factors, and searching optimum conditions for desirable responses. On the other hand, by this method, the interactions of possible influencing parameters on treatment efficiency can be evaluated with a limited number of planned experiments (Khaloo et al. 2015).
Based on the facts mentioned above, in this study a nano-composite of NiFe2O4-graphene oxide modified by a poly aminoglycoside (neomycin) was synthesized. Sonochemical nucleophilic substitution reaction was applied to modify the graphene oxide surface by neomycin (Ne). NiFe2O4 was synthesized by a simple hydrothermal process and was directly loaded on the modified graphene oxide (MGO). The chemical structure of nano-composite was characterized by different techniques such as scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), X-ray diffraction (XRD) and Fourier transform infrared (FT-IR) spectroscopy. The synthesized magnetic modified nano-composite (MMNC) were used as nano-adsorbent for removal of 4-nitrophenol (4-NP) in water. The adsorption kinetic and isotherm parameters of the adsorption process were investigated. A three-factor Box–Benhken design coupled with response surface method (BBD-RSM) was applied to evaluate and optimize the important experimental variables and a quadratic model was derived using analysis of variance to predict the effects of different parameters and their interactions on the removal efficiency of MMNC.
MATERIALS AND METHOD
Chemicals and reagent
Ferric chloride hexahydrate, nickel chloride hexahydrate, potassium permanganate, graphite powder, sulfuric acid (95%), hydrogen peroxide (30%, w/w) and 4-nitrophenol were purchased from Merck (Darmstadt, Germany). Neomycin sulfate was purchased from Sigma-Aldrich Co. All chemicals were of analytical grade and used as received without further purification. De-ionized (DI) water was used as the solvent throughout the experiments. All of the working solutions used for the optimization and establishing the isotherms were prepared by successive dilution of stock solutions.
Synthesis of modified magnetic nano-composite
GO was synthesized from graphite powder by a modified Hummer's method as published previously (Khaloo et al. 2016). The synthesized GO 5 mg/ml suspension in ethanol was exfoliated with ultrasonic radiation (Elmasonic E 60 H, 37 kHz 400 W, continuous wave mode, Germany) to prepare graphene oxide nanosheets and also to enhance the chemical reactivity of GO. The color of the particles changes to light brown after 2 hours sonication. Ten millilitres of 30 mg/ml neomycin sulfate was added to the light brown particles to start the chemical reaction of imide bonds formation. The reaction took place in 30 min to develop a black particles suspension. The MGO was centrifuged at 4,000 rpm and washed with DI water three times.
The MMNC was synthesized by functionalization of MGO with nickel ferromagnetic particles by the hydrothermal method. In a typical procedure, the as-prepared MGO (450 mg) was exfoliated by sonication in 60 ml of DI water for 15 min. Then, 25 ml aqueous solution containing 1 g of FeCl3.6H2O, 0.44 g of NiCl2.6H2O (molar ratio of Fe:Ni, 2:1) in 0.4 M HCl was dropped into the MGO suspension at 80 °C followed by stirring for 30 min. After that, the pH of the mixture was adjusted to about 11 by dropping in 1 M NaOH. The mixture was transferred into a 100 ml Teflon-lined stainless steel autoclave and heated at 170 °C for 10 h. A solid black product was obtained and washed several times with DI water. The obtained nano-composite was dried at 50 °C overnight.
The morphology and elemental analysis of the synthesized magnetic nano-particles were assessed by obtaining SEM images, recorded using a Czech Republic MIRA3, TESCAN scanning electron microscope equipped with an EDS accessory. The microscopic properties, such as the crystal structure, the crystal orientation and crystal size of the prepared MMNC samples, were recorded by XRD (D8 Advanced Bruker diffractometer), using Cu Kα radiation with a scanning range of 2θ 10–90°. FT-IR spectra of GO and MMNC were measured with FT-IR spectroscopy (Rayleigh WQF-510A, Beijing, China).
Experimental design with RSM
All of the experimental results were expressed as mean ± standard deviations of three duplicate measurements. Statistical analysis was performed by using the Minitab 16 software (Minitab Inc., State College, PA, USA) and a second-order polynomial regression model containing the coefficient of linear, quadratic and interaction terms. In order to test the model significance and suitability, the analysis of variance (ANOVA) was performed for each response. The significances of all terms in the polynomial were statistically analyzed by computing the F-value at a probability p = 0.05.
RESULTS AND DISCUSSION
Characterization of MMNC
The MMNC was characterized by SEM, equipped with an EDS accessory, and X-ray powder diffraction techniques. An overview of the morphology of graphene oxide, neomycin-modified graphene oxide and MMNC has been shown in Figure 1(a)–1(c), respectively.
As shown, the average thickness of the GO flake measured using SEM is ∼2.1 nm and the particle size of NiFe2O4 particles which have been decorated on GO sheets is less than 20 nm. EDS analysis was performed to determine the elemental composition of the synthesized compounds (Figure 1(a1)–(c1)). The EDS spectrum demonstrated the presence of C, O, N, Ni, and Fe in the structure of MMNC. The C element comes from both graphene nanosheets and neomycin, N comes from neomycin, and O is attributed to both GO and NiFe2O4 nano-particles n and Si element was present from the plated element for SEM measurement.
The X-ray powder diffraction patterns of the GO and NiFe2O4-Ne-GO are shown in Figure 2. A sharp diffraction peak at around 2θ = 10.2o, which corresponds to (001) plane of GO, and no obvious peak of graphite indicate the formation of GO from graphite. This sharp peak was clearly observed in both XRD patterns. The characteristic peaks at 2θ = 18.7, 30.7, 35.5, 38.8, 44.8, 55.6 and 64.3° can be indexed as the (111), (220), (311), (400), (422), (511) and (440) crystal planes of NiFe2O4, respectively. The XRD pattern of MMNC shows a sharp peak at around 10 that corresponds to the plane of GO, which confirmed that the nano-particles of NiFe2O4 coated on MGO and MMNC composite were prepared successfully.
FT-IR spectra of graphene oxide, and MMNC have been illustrated in Figure 2(b). C-O stretching vibration of epoxy and alkoxy appeared at 1,046 and 1,220 cm−1 and the stretching vibration bands of C=C and C=O were observed at 1,600–1,603 and 1,720 cm−1, respectively. The peaks at 3,100–3,500 cm−1 can be attributed to the O-H stretching vibration of carboxyl groups and the adsorbed water molecules and N-H stretching vibration of the amine group of neomycin. The characteristic band at 630 cm−1 is assigned to the intrinsic vibration of nickel ferrite and demonstrates that NiFe2O4 is present and supported on GO in MMNC. The FT-IR spectrum of MMNC shows bands around 2,900 and 2,890 cm−1, which can be attributed to the C-H and C-N stretching vibration. The observed peaks at 1,380 and 1,090 cm−1 correspond to the O-H and C-O stretching vibration of alcohol groups and the peaks at 1,275 and 1,150 cm−1 correspond to the C-O and C-N stretching vibration of ether and amine groups of neomycin, respectively.
Optimization by response surface method
The relationship between the mean square of the model and the error is expressed by the F test and p-value. Table 4 presents the results of the ANOVA for the second-order polynomial model. At a confidence level of 95%, the p-value for the model, linear, quadratic and interaction terms have a signiﬁcant effect (p ≤ 0.05). The high values of statistical parameters R2 (0.9752) and R2adj (0.9593) signify the good fitting of experimental results to the model. The absence of any lack of ﬁt (p > 0.05) also strengthened the reliability of the model. In other words, the proposed model can significantly predict many changes at the level of the response of the regression equation.
In order to improve and analyze multivariate systems, geometric displays, especially two-dimensional and three-dimensional diagrams, are the best. Therefore, response surface graphs (three-dimensional diagrams) for illustrating response (% removal) versus dependent variables (pH, NP, and MMNC) as a function of two factors when the third factor was kept constant at a center point have been shown in Figures 3–5. Figure 3 shows the effect of pollutant concentration and adsorbent dose on the removal efficiency of 4-nitrophenol from aqueous solutions. As is evident from the figure, by increasing the adsorbent dose, the removal efficiency is reduced, and the removal efficiency is reduced by increasing the contaminant concentration. The graph also shows the fact that the two parameters of the concentration of the contaminant and the adsorbent dose interact with each other and the efficiency of the elimination is a function of both parameters. In the range of 30–60 mg/l of the contaminant, the removal efficiency is greater than 90%, with an adsorbent dose of at least 3.0 g/l. As expected, with increasing concentrations of pollutants, since most surface sites are occupied by the pollutant, the efficiency of removal is reduced and, in order to maintain the efficiency of removal, it is necessary to increase the amount of MMNC simultaneously. On the other hand, by using Figure 3(b) the best dosage of adsorbent could be selected according to the pollutant concentration and the desired removal efficiency.
The response surface and the corresponding contour plots of the 4-nitrophenol removal efficiency have been shown against the contaminant concentration and pH of the solution in Figure 4. As is evident from the figure, by increasing the pH of the solution from 5 to 6.5 in the concentration range of 30 to 50 mg/l, the removal efficiency increases and then, by increasing pH, the removal efficiency is reduced. As expected from the chemical structure of the pollutant, with increasing pH, de-protonation of contaminants increases and therefore the negative charge increases. On the other hand, the adsorbent surface possesses a positive charge in acidic and a negative charge in alkaline environments. Therefore, electrostatic repulsion prevents adsorption of 4-NP in alkaline environments and reduces the removal efficiency.
The US EPA has established for 4-nitrophenol the criterion of less than 550 μg/l to protect freshwater aquatic life (US EPA 1980). As Figure 4 shows, the residual level of NP after adsorption will be less than the above-mentioned criterion when its initial concentration is less than 45 mg/l, pH is 5–6.7 and the minimum dose of MMNC is 3.0 g/l. This result confirms the high efficiency of the synthesized MMNC in the adsorptive removal of 4-NP.
According to Figure 5 by increasing the adsorbent dose in the range of 3.5 to 4 g/l and pH from 5.6 to 6.6, the removal efficiency higher than 95% is achieved. Thus, at all concentrations, the best removal efficiency is achieved at pH 6. By increasing the acidity of the solution, as a result of the simultaneous protonation of the adsorbent and the contaminant, the electrostatic repulsion increases and consequently the removal efficiency is reduced. At pH values higher than pKa of 4-NP, the same trend of increasing negative charge resulting from de-protonation of contaminants and adsorbent surface reduces removal efficiency. Such a trend of surface charge changes is only pH-dependent and is present in all adsorbent dosages. Therefore, the dosage of adsorbent and pH of the environment do not have a significant effect on each other. The results of the ANOVA in Table 3 also confirm the insignificant interaction of these two independent parameters (MMNC*pH p-value is 0.954).
The optimum parameters for high removal efficiency of 4-NP as specified by RSM are initial contaminant concentration 60 mg/l, pH 6 and adsorbent dose 3 g/l. Based on the predictive model, removal efficiency of 4-NP at the optimum condition was calculated as 94.8%. Experimental removal of 93.3% matches quite well with the predicted one, which validated the RSM models with good correlations (R2 = 0.9897). On the other side, response surface graphs and their corresponding contour help to find the optimum parameters to get the desired removal efficiency according to the other parameters.
Isotherm study of the adsorption process
In the above equations, C0 and Ce denote the initial and equilibrium concentration of the adsorbate (mg/l), respectively. qt and qe are the temporary and equilibrium loading of sorbate on sorbent (mg/g), qm and b are the Langmuir constants related to the maximum adsorption capacity (mg/g), and the relative energy of adsorption (l/mg), respectively. Kf and n are the Freundlich constants related to adsorption capacity and adsorption intensity, respectively. V is the volume of the solution (l) and m is the weight of sorbent (g).
Figure 6 shows the adsorption isotherm of the process using Langmuir and Freundlich models. The isotherm parameters which have been derived from both models have been given in Table 5. As can be concluded from data of the table, the most precise match was acquired using the Langmuir model. Hence, this confirmed that the adsorption of 4-NP is a monolayer adsorption process on a surface containing a finite number of identical sites. Table 5 also shows that the presence of neomycin on the surface of the magnetic nano-particle improves the maximum absorbance capacity by up to 28%. The maximum adsorptive capacity of magnetic modified graphene oxide was found to be 125.6 mg of NP/g sorbent which was comparable or higher in some cases to the other reported sorbents for removal of NP (Table 6). Neomycin provides several amine, alcohol and ether groups at the surface of GO and then different interactions, such as electrostatic interaction and hydrogen bonding, could improve adsorption efficiency of 4-NP on the surface of MMNC (Lu et al. 2017).
Kinetic study of the adsorption process
In the above equations, qe and qt denote the sorption capacity (mg/g) at equilibrium and time t, respectively. K1 (min−1) and K2 (gg/(mg·min)) are the pseudo-first- and the pseudo-second-order constants. Figure 7 illustrates the kinetic studies of 4-NP adsorption on MMNC according to the above-mentioned equations. The kinetic parameters that were evaluated from the slope and intercepts of the graphs are summarized in Table 7. The values of the correlation coefficients proved that the pseudo-second-order adsorption model is more suitable to describe the adsorption kinetics of 4-NP on the surface of MMNC.
In this study, NiFe2O4-Ne-GO was synthesized as a new nano-composite adsorbent. The synthesized nano-adsorbent was characterized by SEM, EDS and XRD techniques. The applicability of the synthesized nano-composite was assessed for adsorption removal of 4-nitrophenol as a representative of the phenolic pollutants. The adsorption process has fast kinetics so that more than 95% of absorption occurs in the first 40 minutes of contact. RSM coupled with BBD was used for the optimization of the operation parameters for the adsorption removal of 4-NP. Response surface curves showed that the synthesized MMNC in the removal of 4-NP has excellent adsorption capability in such a way that the residual level of 4-NP after adsorption will be less than the criterion to protect freshwater aquatic life (550 μg/l). Thus, this adsorptive system can be introduced as a new sorbent for the removal of phenolic compounds from aqueous environments in mixing tank reactors, continuous-flow or intermittent flow.
The authors gratefully acknowledge the support of this work by the Shahid Beheshti University of Medical Sciences.
CONFLICT OF INTEREST
The authors have declared no conflict of interest.