Modeling of azo dyes adsorption on magnetic NiFe₂O₄/RGO nanocomposite using response surface methodology

Abstract

Background

Azo group dyes are the largest group of synthetics dyes that widely used in industries, especially in textile industry. The presence of these organic compounds in wastewaters and their discharge into environment without efficient treatment may cause adverse effect on human, living and aquatic environment. The purpose of this study was to optimize the adsorption of azo dye of Direct Red 81 (anionic dye) and Basic Blue 41 (cationic dye) from aqueous solution onto magnetic NiFe₂O₄/RGO nanocomposite.

Methods

In this study the response surface methodology (RSM) based on the central composite design (CCD), was used to optimization and modeling of adsorption process DR81 and BB41 dye on NiFe₂O₄/RGO. in order to investigating the effect of the operating parameters on the adsorption efficiency DR81 and BB41, four influential factors were chosen that includes of pH (3–9), contact time (5–25 min), adsorbent amount (0.02–0.05 g) and initial dye concentration (40–200 mg/L). A total of 30 experiments were performed for each dye in this study. The concentration of dye in solution was measured by spectrophotometer. The structure of synthesized adsorbent was investigated using Scanning Electron Microscope (SEM), X-ray diffraction (XRD), Fourier transform irradiation (FTIR), transmission electron microscope (TEM) and vibrating sample magnetometer (VSM).

Results

Analysis of variance (ANOVA) showed that regression model for both dye adsorption with value of P value <0.001 is significant statistically. The correlation coefficient (R²) for DR81 (R² = 0.9968) and BB41 (R² = 0.9948) indicated which there is a good agreement between predicted values and the results of the experiments and the model also well predict the adsorption efficiency. Furthermore, the factors of pH, dye concentration and adsorbent dose, have the greatest effect on adsorption, respectively, while contact time have the lowest effect on adsorption of both dyes. The adsorption behavior of the DR81 and BB41 onto NiFe₂O₄/RGO was best described by the Langmuir and Freundlich isotherm, respectively. The optimum conditions for maximum removal of DR81 (96.41%) was found to be at pH 3, contact time 19.68 min, adsorbent dose 0.02 g and initial dye concentration 40 mg/L. However, the optimum conditions for maximum removal of BB41 (97.87%) was found to be at pH 9 contact time 18.16 min, adsorbent dose 0.02 g and initial dye concentration 40 mg/L.

Conclusion

The present study shows that magnetic NiFe₂O₄/RGO nanocomposite have much potential as a powerful adsorbent for the rapid adsorption of anionic (DR81) and cationic dyes (BB41) from aqueous solution.

Introduction

Dye are organic compounds which are used extensively in industries such as textiles, printing, paper, leather, rubber, food, cosmetics, plastics, pharmaceuticals, photography and etc. [1]. It is estimate that more than 100,000 various dye types and 700,000 tons of dyes are used annually in different industries which during dyeing operation about 10 to 15% of the total applied dyes discharge into sewer. Azo group dyes which are characterized with one or more azo bonds (–N=N–), is the largest group of dyes used in industries in the world, especially in textile industry [2, 3]. These dyes are known as pollutants in the environment and they are non- degradable, toxic, carcinogenic, teratogenic or mutagenic to human and living due to the presence of benzene rings in their aromatic structure [4], and have adverse effect on aquatic environment because of preventing the penetration of sunlight into water, precluding the photosynthetic activity of aquatic plants and decrease dissolved oxygen [5].

Many methods have been used for the treatment of dye contaminated industrial wastewater including chemical oxidation [6], photocatalytic degradation [7], biological treatment [8], Chemical coagulation/flocculation [9], ozonation [10], enzymatic treatment [11–13], ion exchange [14] and adsorption process [15]. Adsorption is an economic and effective technique for removal of dye pollutants from wastewater and in comparison, with other processes have several advantages such as low cost, simplicity in design, easy operation, insensitivity to toxic material, high efficiency, and ability to treat concentrated effluents [16, 17] .In general wide range of adsorbents such as activated carbon [18], agricultural waste [19], chitosan [20], zeolites [21], clays [22], polymeric adsorbents [23], perlite [24], kaolin [25] and etc. have been used for the removal of dyes from industrial effluents. However, researches always try to find a novel adsorbent with high performance for removal of dye from wastewater. Nowadays, nanostructured adsorbents due to large specific surface and high adsorption capacity have been much attention for wastewater treatment. Graphene as one of the carbon allotropes is a flat single-layer from graphite with sp2 hybridized carbon atoms covalently bonded, that atoms arranged in a two-dimensional honeycomb network [26]. In recent years the graphene have attracted much attention in science owing to its unique properties such as electrical, optical, mechanical, high thermal conductivity and large specific surface area [27, 28]. Because of this extraordinary properties, it has wide applications in environmental pollution purification and various of industries such as electronic devices, solar cells, sensors, batteries, supercapacitors, hydrogen storage [29, 30] .Graphene, with high specific surface area, the large delocalized π-electron system and good chemical stability can make a strong π–π interactions with the benzene rings present in aromatic structure of dyes [31], therefore, it can be used as an effective and high-performance absorbent for removal of colored pollutant from aqueous solutions. One of the best ways to producing graphene in large scale, low cost and high-yield production is reduction of graphene oxide (GO) to reduced graphene oxide (RGO) by chemical method [32]. The mechanism of dye molecules adsorption on RGO surface is using of electrostatic forces, oxygen functional groups, vacancy defects and π–π interactions [33]. The separation of graphene-based absorbents from solution after adsorption process is very difficult due to their small particle size. Therefore, in order to solve this problem, recently many researchers have been focused on the magnetic graphene-based material that facilitate separation from solution rapidly and effectively by magnetic field [34]. For example Hou Wang and et al. reported the facile synthesis of polypyrrole decorated reduced graphene oxide-Fe₃O₄ magnetic composites as absorbent for the removal of Cr(VI) [35]. Yonrapach and et al. investigated the enhanced sonocatalytic degradation of organic dyes from aqueous solutions by novel synthesis of mesoporous Fe₃O₄-graphene/ZnO@SiO₂ nanocomposites [36].

In recent years, magnetic ferrite nanocrystals with the general formula MFe₂O₄ (M = Fe, Co, Cu, Mn, Ni, etc.) have attracted much attention in fields of technology and environmental remediation owing to considerably sized and shape magnetic nanostructures properties. Nickel ferrite (NiFe₂O₄) with an inverse spinel structure is one of the important ferrites which used as a soft magnetic material and has a ferromagnetic behavior that originates from a magnetic moment of antiparallel spins between Fe³⁺ ions at tetrahedral sites and Ni²⁺ ions at octahedral sites [37, 38]. this ferrite is a magnetic material with high-electrical resistivity, high surface area high-Curie temperature and environmental stability that widely used in power and signal transformation, storage devices, microwave devices, gas sensors, ferrofluids, and catalysts [38, 39]. Considering the unique properties of graphene and superparamagnetic property of NiFe2O4, the combination of them, can be used in form of magnetic NiFe₂O₄/RGO nanocomposite as an efficient and recoverable adsorbent to removal of contaminants, such as synthetic dyes in wastewater treatment. The separation this nanocomposite of solution during the adsorption process is easy by a magnet due to it magnetic property.

The response surface methodology (RSM) is a combination of statistical and mathematical techniques which widely used in analytical chemistry for determine of main and interaction effects of independent variables (input) on the response variable (output) in the any process and also determine of factors that have more effect on process [40, 41]. This statistical method establishes an empirical models based on the data obtained from experimental results in order to evaluate the relationship between independent factors and responses as well as the to optimize the optimum condition [42]. Another objective of this method is determine optimal levels of operating parameters in order to achieve the best performance in process [43]. Among several matrix designs in RSM, central composite design (CCD) due to the simple structure and high efficiency is most commonly method applied to investigate individual and combined effects of independent parameters on response with least number of experiments [44, 45]. The present work was to investigate the adsorption of Direct Red 81 (anionic dye) and Basic Blue 41(cationic dye) as a model of azo dyes from aqueous solutions using magnetic NiFe₂O₄/RGO nanocomposite as an adsorbent by applying response surface methodology (RSM).

Materials and methods

Materials

Direct red 81 (C₂₉H₁₉N₅Na₂O₈S₂) is a diazo dye with molecular weight of 675.59 g/mol and the Basic blue 41 (C₂₀H₂₆N₄O₆S₂) is a mono azo dye with molecular weight of 482.57 g/mol. Both synthetic dyes were of analytical grade and were provided by Alvan Sabet Co. (Tehran, Iran). All other chemicals were purchased from Sigma-Aldrich and used without further purification.

Adsorbent characterization

XRD patterns were recorded on a Philips X’pert diffractometer with mono chromatized Cu Kα radiation at 40 kV in the range of 2θ 10-80^o. SEM imaging was performed on an EM-3200 scanning electron microscope. FT-IR spectra were recorded in KBr disks on an ALPHA Bruker instrument. TEM images were recorded on a MC 10 transmission electron microscope from PHILIPS with an acceleration voltage of 80 kV. VSM magnetization curves were recorded on a LBKFB instrument from Meghnatis Daghigh Kavir Company, Iran.

Preparation of magnetic NiFe₂O₄/RGO nanocomposite

The approach of the preparation of NiFe₂O₄/RGO includes four steps comprising the preparation of the GO, the preparation of the NiFe₂O₄ nanoparticles, anchoring the NiFe₂O₄ nanoparticles onto the GO structures and finally reducing the NiFe₂O₄/GO to NiFe₂O₄/RGO [46].

Graphene oxide was prepared according to the method of Marcano et al. [47]. In brief, graphite (3.0 g) was added to a mixture of concentrated H2SO4/H3PO4 (360:40 mL), and then 18.0 g of KMnO4 was added. The reaction was then stirred at 50 °C for 12 h. After cooling to room temperature, the mixture was poured into a mixture of ice (400 g) and hydrogen peroxide (3.0 mL, 30%). The solid was then filtered through polyester fiber (Carpenter Co.) and the filtrate centrifuged at 4000 rpm for 4 h. The precipitate was then washed with water, HCl (30%), and ethanol (200 mL of each), and vacuum-dried overnight.

Nickel ferrite (NiFe₂O₄) nanoparticles were synthesized via a modified co-precipitation method [48]. An aqueous solution of Fe (NO₃)₃.9H₂O and Ni (NO₃)₂.6H₂O was prepared in 2:1 M ratio. A specified amount of oleic acid was added to the solution as the surfactant. The solution was heated to 90 °C. In the following, the pH was adjusted to 10 by 28% ammonia solution while vigorously stirred. The liquid precipitate was then brought to a reaction temperature of 80 °C and stirred for 30 min. The product was cooled to the room temperature and then washed twice with distilled water and ethanol to eliminate the unwanted impurities and the residual surfactant from the prepared sample. To the next step, this product was re-dispersed in distilled water using bath sonication for 15 min. In the following steps: at first, 5 g of GO was dispersed in 200 mL distilled water using an ultrasonic probe (FAPAN 400UT- FAPAN Ultrasound Technology, Iran) for 1 h, leading to the formation of a GO brown suspension. This suspension was added to the ferrite nanoparticles’ suspension in a 1:1 mass ratio while vigorously stirred. Resulting suspension was sonicated for 30 min using probe sonicator. Finally, the suspension was heated to 90 °C. The hydrazine hydrate was added with vigorous stirring to the mentioned suspension. After stirring for 2 h, the dark-brown suspension was centrifuged, washed with water and ethanol several times.

Batch adsorption studies

The stock solution of Direct Red81 and Basic Blue 41 was prepared at a concentration of 1000 mg/L by dissolving in deionized water. The required dye concentration was prepared from the stock solution by dilution. In order to study the adsorption process by NiFe₂O₄/RGO the adsorption experiments were carried out by batch method. The influence of various operating variables on the adsorption including pH, contact time, dye initial concentration and adsorbent dose were investigated. The experiments were performed by mixing various amounts of NiFe2O4/RGO in a 100 mL beaker containing 50 mL dye solution (DR81 and 41) of known concentration. The pH of the solution was adjusted to the desired value using 0.1 N NaOH and 0.1 N HCl and was measured by a pH meter (met Rohm, 827 pH lab). Then the mixture was agitated at a speed of 180 rpm by a stirrer for a predefined time. The chemical structure of mentioned dyes is shown in Fig. 1.

Fig. 1.

Chemical structure of dyes DR81 (a) and BB41 (b)

At the end of adsorption process, the NiFe₂O₄/RGO was separated from the solution using a magnetic field and the residual concentration of the dye in the solution was measured by a spectrophotometer (UV DR5000) at a maximum absorbency wavelength (λmax) of 510 nm for DR81 and 609 nm for BB41. The adsorption percentage of dye was calculated using the following equation:

$$R\%=\frac{C_0-C_t}{C_0}$$ 1

Where, Co (mg/L) and Ct (mg/L) are the initial and equilibrium concentration of dyes in aqueous solution, respectively.

Experimental design with response surface methodology (RSM)

In this study, the response surface methodology (RSM) based on the central composite design (CCD), one of the most popular methods of designs, was used for the optimization and modeling of the adsorption process of DR81 and BB41 dye on NiFe₂O₄/RGO. Therefore, in order to investigate the effect of the operating parameters on the adsorption efficiency of DR81 and BB41, as shown in Table 1, four main and influential factors were chosen, which included pH, contact time, adsorbent amount and initial dye concentration in three coded levels (−1, 0, +1,). A total of 30 experiments were performed for each dye in this study including, 16 cube points, 6 replicates at the center point and 8 axial points. The number of experimental runs was determined by the following equation:

$$\mathrm{N}=2^{\mathrm{n}}+\text{2n}+{\mathrm{n}}_{\mathrm{c}}$$ 2

Table 1.

Independent variables and levels used for response surface design

Independent Factors	Coded symbol	Level
		Low (−1)	Central (0)	High (+1)
pH	A	3	6	9
Contact time (minute)	B	5	15	25
Adsorbent dose (g)	C	0.02	0.035	0.05
Initial dye concentration(mg/L)	D	40	120	200

Where N is the total number of experiments, n is the number of operational factors, and n_c is the number of replications in central points.

An empirical second order quadratic equation a model was used to evaluate the interaction between the independent (input) and the dependent (output) variables expressed with the following equation:

$$\mathrm{Y}={\mathrm{\beta }}_0+{\sum }_{i=1}^K{\beta }_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i}}+{\sum }_{i=1}^K{\beta }_{\mathrm{ii}}{{\mathrm{X}}_{\mathrm{i}}}^2+{\sum }_{i=1}^{n-1}a{\sum }_{j=2}^n{\beta }_{\mathrm{ij}}\mathrm{Xi}\mathrm{Xj}$$ 3

Where y expresses the response or adsorption percentage, β0 indicates the intercept constant, Xi and Xj represent independent variables, βi shows the coefficients of linear terms, βii represents the coefficients of quadratic terms, βij shows the coefficients of interaction terms, ε indicates experimental error and k represents the number of independent variables.

The analysis of variance (ANOVA) was used in order to evaluate the validity, significance and accuracy of model and also statistical significance of effect of each variable. Statistical significance of quadratic model is determined using lack-of-fit, correlation coefficient (R²), adjusted correlation coefficient (R² adj) between predicted and experimental results. The Design Expert software version 7.0.0 (Stat-Ease, trial version) was utilized to analyse and interpret the results of the experimental design.

Results and discussion

Characterization of the NiFe₂O₄/RGO

XRD patterns of GO, NiFe₂O₄, and NiFe₂O₄/RGO are compared in Fig. 2. In the XRD pattern of the composite, corresponding reflections of single phase cubic NiFe2O4 nanoparticles are clearly visible at 2θ 30.5, 35.9, 43.4, 53.7, 57.4, 62.9, 74.4, and 75.7 according to standard values in the card (JCPDS card No. 074–2081). The index peak of GO at 2θ 12, was eliminated in the XRD pattern of the composite which can be an evidence of reduction to RGO.

Fig. 2.

XRD patterns of GO (a), NiFe₂O₄ (b), and NiFe₂O₄/RGO (c)

FTIR spectra of NiFe2O4, GO and NiFe₂O₄/RGO are shown in Fig. 3. It is clear that in the FTIR spectrum of NiFe₂O₄/RGO, the characteristic vibration of C=O functional groups of GO at 1740 cm-1 has been disappeared. Broad band at about 3450 cm-1 is attributable to the O-H stretching of the adsorbed water molecules, and C=C stretching vibrations are located at about 1640 cm-1. Metal-oxygen vibrations are also clearly visible at about 600 cm-1.

Fig. 3.

FTIR spectra of NiFe₂O₄ (a), GO (b) and NiFe₂O₄/ RGO (c)

In order to gain an insight to the morphology of the synthesized materials, scanning electron microscopy (SEM) was used. Figure 4 shows the micrographs obtained for GO, NiFe₂O₄ and NiFe₂O₄/RGO. GO was appeared as non-geometric sheets. Nickel ferrite nanoparticles were appeared as many ultrafine spherical particles which were aggregated due to their intrinsic magnetic properties. Interestingly, NiFe₂O₄/RGO composite was formed as an almost uniform dispersion of NiFe₂O₄ nano-spheres over the surface of reduced graphene oxide sheets.

Fig. 4.

SEM images of GO (a), NiFe₂O₄ (b) and NiFe2O4/ RGO (c)

To make graphene sheets appear more clearly, TEM imaging was performed. In Fig. 5, part of an uncoated sheet is seen at the bottom. However, at the top of the image, the dense nanoparticles aggregated on the graphene sheet prevent it from being seen.

Fig. 5.

TEM image of NiFe₂O₄/ RGO

Figure 6 Shows photographs of dye solution DR81 and BB41 before and after adsorption using NiFe₂O₄/RGO and effect of magnetic separation. As shown, after shaking the mixture of NiFe₂O₄/RGO and dye solution, nanocomposite can be good dispersed in the aqueous solution and forming black suspension. After adsorption, NiFe₂O₄/RGO nanocomposites can be easily separated from the solution in a short time (30s) by placing an external magnetic close to beaker and at the result the solution became transparent. This demonstrated that the NiFe₂O₄/RGO nanocomposite with having the properties of adsorption capacity and magnetism, could be used as a magnetic and effective adsorbent to remove organic pollutants in aqueous solution. Magnetic behavior of NiFe₂O₄/ RGO as a function of applied magnetic field at 298 K is shown in Fig. 7. The NiFe₂O₄/ RGO sample exhibited superparamagnetic behavior with a large saturation magnetization of 31.0 emu.g⁻¹

Fig. 6.

Photographs of magnetic separation of NiFe₂O₄/ RGO adsorbent from solution with external magnetic field (a; DR81 and b; BB41)

Fig. 7.

VSM magnetization cure for NiFe2O4/ RGO measured at 298 K

Statistical analysis and analysis of variance (ANOVA)

By applying the analysis of multivariable regression on experimental data (Table 2), the relation of response (adsorption efficiency) and the investigated variables was obtained using the quadratic equation (Eq.1) according to the following equations for DR81 and BB41 respectively:

$$\%\mathrm{R}\mathrm{DR}81=43.62–23.78\mathrm{A}+1.24\mathrm{B}+6.74\mathrm{C}–11.25\mathrm{D}+0.32\mathrm{AB}–0.084\mathrm{AC}+2.50\mathrm{AD}+0.098\mathrm{BC}+0.23\mathrm{BD}+3.99\mathrm{CD}+22.77{\mathrm{A}}^2-0.87{\mathrm{B}}^2-2.21{\mathrm{C}}^2-1.56{\mathrm{D}}^2$$ 4

$$\%\mathrm{R}\mathrm{BB}41=65.36+22.70\mathrm{A}+1.50\mathrm{B}+6.38\mathrm{C}–10.18\mathrm{D}+0.073\mathrm{AB}+0.58\mathrm{AC}–2.03\mathrm{AD}+0.20\mathrm{BC}+0.47\mathrm{BD}+3.51\mathrm{CD}+1.47{\mathrm{A}}^2–2.07{\mathrm{B}}^2–1.62{\mathrm{C}}^2–0.91{\mathrm{D}}^2$$ 5

Table 2.

Experimental and predicted results and variable matrix from CCD for DR81 and BB41

Independent factors					Adsorption efficiency (%)
					DR81		BB41
Run	A	B	C	D	Experimental	Predict	Experimental	Predict
1	9	25	0.05	200	39.67	40.76	86.86	85.03
2	6	5	0.035	120	40.88	41.50	61.70	61.80
3	9	5	0.05	200	35.24	36.97	82.07	81.36
4	9	5	0.02	200	18.55	15.87	58.41	60
5	3	5	0.02	40	96.12	94.86	42.25	44.23
6	6	15	0.035	120	43.50	43.62	65.69	65.36
7	6	15	0.035	120	44.80	43.62	64.50	65.36
8	6	15	0.035	120	42.06	43.62	65.03	65.36
9	3	5	0.05	200	81.04	80.35	37.62	39.01
10	6	15	0.02	120	32.79	34.67	58.06	57.36
11	3	25	0.02	200	59.62	61	24.67	24.17
12	9	25	0.02	200	20.24	19.26	62.09	64.48
13	6	15	0.035	40	53.83	53.31	74.80	74.63
14	3	25	0.05	40	99.40	101.90	52.17	50.73
15	9	5	0.02	40	40.12	41.81	94.23	92.37
16	6	15	0.05	120	49.89	48.15	71.41	70.12
17	3	5	0.05	40	99.20	100.32	50.45	49.22
18	6	15	0.035	120	44.22	43.62	64.05	65.36
19	9	25	0.02	40	43.78	44.30	96.23	94.99
20	3	25	0.02	40	97.64	96.05	45.50	46.56
21	9	15	0.035	120	40.52	41.61	89.96	89.53
22	6	15	0.035	120	43.11	43.62	64.73	65.36
23	6	25	0.035	120	44.48	43.99	66.88	64.79
24	3	5	0.02	200	57.87	58.91	21.87	19.97
25	6	15	0.035	120	44.45	43.62	62.2	65.36
26	9	25	0.05	40	50.71	49.81	99.24	101.49
27	3	15	0.035	120	90.13	89.18	45.70	44.14
28	9	5	0.05	40	48.49	46.93	99.04	99.69
29	3	25	0.05	200	84.38	82.84	40.17	42.38
30	6	15	0.035	200	30.16	30.81	56.09	54.27

In these eqs. R is the adsorption efficiency percent, A, B, C and D are the independent variables pH, contact time (minute), adsorbent dosage (g/l) and initial dye concentration (mg/l), respectively.

Analysis of variance (ANOVA) was used to investigate the accuracy and significance of the statistical model and also statistical significance of individual and interaction effects of each variable on adsorption process. Statistical significance of the model was checked by P- value, lack of fit and R². A P- value less than 0.05 demonstrates the effect of parameters and model items are significant (at 95% confidence level) [49].

According to the results presented in Tables 3 and 4, the quadratic model with P value <0.0001, F-value = 337.32 for DR81 and P value <0.0001, F-value = 204.41 for BB41 is statistically significant. The lack of fit of the model for both dyes was insignificant (p value = 0.0551, F-value = 4.51 for DR81 and p value = 0.0696, F-value = 4 for BB41), it implies that the proposed model has a good fit to the experimental data and the independent variables have significant effects on the response. The R² values of model for DR81 (0.9968) and BB41 (0.9948) that indicated suitability more than 99.68% and 99.48% of the data obtained from the experiments with the predict data by the model for DR81 and BB41, respectively. On the other hand, the adjusted R² (0.9939) for DR81 and (0.9899) BB41 was close to R² (0.9968 and 0.9948) that confirmed there is good agreement between the predicted values by the model and results of the experiments for both dye and implied that the proposed model can well predict experimental results. Also, the low value of the Coefficient of Variation (3.43 and 3.26% for DR81and BB41, respectively) indicates the accuracy of the measurements and the reliability of the experiments. The adequate precision measures signal to noise ratio and a ratio greater than 4 is desirable. Adequate precision for adsorption model of DR81 (65.790) and BB41 (55.709) demonstrating an adequate signal, therefore model can be used to navigate the design space.

Table 3.

ANOVA for adsorption of DR81

Source	Sum of squares	Degree of freedom	Mean square	F-Value	Prob > F
Model	16,151.68	14	1153.69	337.32	<0.0001
A	10,180.69	1	10,180.69	2976.65	<0.0001
B	27.90	1	27.90	8.16	0.0120
C	817.29	1	817.29	238.96	<0.0001
D	2278.58	1	2278.58	666.21	<0.0001
AB	1.68	1	1.68	0.049	0.4937
AC	0.11	1	0.11	0.033	0.8576
AD	100.25	1	100.25	29.31	<0.0001
BC	0.15	1	0.15	0.045	0.8348
BD	0.81	1	0.81	0.24	0.6326
CD	255.28	1	255.28	74.64	<0.0001
A2	1228.03	1	1228.03	359.05	<0.0001
B2	1.98	1	1.98	0.58	0.4585
C2	12.70	1	12.70	3.71	0.0731
D2	6.30	1	6.30	1.84	0.1949
residual	51.30	15	3.42
Lack of fit	46.18	10	4.62	4.51	0.0551
Pure error	5.12	5	1.02

SD 1.85, Mean 53.90, CV 3.43, AP 65.79, Press 354.35, R² 0.9968, R²(adj.) 0.9939

Table 4.

ANOVA for adsorption of BB41

Source	Sum of squares	Degree of freedom	Mean square	F-Value	Prob > F
Model	12,255.74	14	875.41	204.41	<0.0001
A	9272.50	1	9272.50	2165.16	<0.0001
B	40.44	1	40.44	9.44	0.0077
C	733.57	1	733.57	171.29	<0.0001
D	1865.59	1	1865.59	435.62	<0.0001
AB	0.084	1	0.084	0.020	0.8904
AC	5.43	1	5.43	1.27	0.2779
AD	65.85	1	65.85	15.38	0.0014
BC	0.67	1	0.67	0.16	0.6975
BD	3.48	1	3.48	0.81	0.3817
CD	197.26	1	197.26	46.06	<0.0001
A2	5.61	1	5.61	1.31	0.2703
B2	11.08	1	11.08	2.59	0.1285
C2	6.83	1	6.83	1.59	0.2260
D2	2.16	1	2.16	0.50	0.4884
Residuals	64.24	15	4.28
Lack of fit	57.10	10	5.71	4	0.0696
Pure error	7.14	5	1.43

SD 2.07, Mean 63.48, CV 3.26, AP 55.709, Press 394.97, R² 0.9948, R²(adj.) 0.9899

As shown in Tables 3 and 4, according to statistical analysis of studied variables on adsorption process, it can be realized that all variables have significant effects on the adsorption efficiency of both dye by NiFe₂O₄/RGO (P value <0.05), with the explanation that pH have the greatest effect on the adsorption efficiency of both dye (F-value of 2976.65 and 2165.16 for DR81 and BB41 adsorption, respectively). Although, the contact time with the less value of F-value has the less effect than three other variables on the adsorption efficiency (F-value of 8.16 and 9.44 for DR81 and BB41 adsorption, respectively). On the other hand, the interaction effects of variables of AD and CD for both dye are statistically significant.

To evaluate the accuracy and adequacy of the model, the difference between predicted and residual values was used to check the accuracy of the model graphically. The residuals are considered as non-fitted changes by the model. Figure 8a and b, depicts the normal probability plot versus residuals with a 95% confidence for adsorption of studied dyes. Normal probability plot is a graphical tool for testing whether the data follows a normal distribution or not. If the data points on the normal probability plot of the residues form an approximate straight line, indicated that the data are normally distributed and departure from this straight line indicates departure from normality. From this figure, it is confirmed that the data points form the approximately straight line, thus indicating the data set is approximately normally distributed and data are reliable [19].

Fig. 8.

Normal plots of residuals for DR81 (a) and BB41 (b), Plot of the experimental versus predicted data of DR81(c) and BB41 (d) adsorption

The plot of relationship and correlation between predicted values by model with actual values of the dye adsorption percentages at 95% confidence interval was shown in Fig. 8 (c) and (d), for DR81 and BB41, respectively. It is obvious from the plot that the predicted data for both dye adsorption processes were close to the actual data. On the other hand, the values of R² and adjusted R² for plot of DR81 (R²; 0.9968, Adj. R²; 0.9939) and BB41 (R²; 0.9948, Adj. R²; 0.9899), indicated there is a good agreement between the predicted values and the results of the experiments and also significance of the model was applied for the adsorption of DR81 and BB41 on NiFe₂O₄/RGO. Therefor the quadratic model is suitable for predicting the adsorption efficiency of both dye in the studied conditions.

Effect of interactive variables and 3D response surface plot

Figures 9a–f and 10a–f shows the three-dimensional (3D) plots of the effect of the all independent variables on the adsorption efficiency of DR81 and BB41 dyes, respectively. In the case of pH effect on the adsorption efficiency of DR81, (Fig. 9a–c), can be realized that with decreasing of pH from 9 to 3 the adsorption efficiency has increased significantly. This is due to that in highly acidic conditions (pH = 3), H⁺ ions increases in the solution as well as the adsorbent surface is protonated and positively charged. Consequently, anionic molecules of DR81 with negatively charge is adsorbed on the positive surface of NiFe₂O₄/RGO through electrostatic force. On the other hand, in alkaline conditions with the increasing of pH up to 9, OH⁻ ions increased in the solution so the adsorbent surface is negatively charged and consequently the electrostatic force and adsorption efficiency decreased. In the case of BB41 adsorption efficiency, it is noticeable that with increasing of pH from 3 to 9, the adsorption efficiency has increased significantly. This increases trend for adsorption is due to the increasing of OH⁻ ions in the solution, which causes the surface negatively charged of adsorbent consequently the cationic molecules of BB41 with positively charge is adsorbed on the negative surface of NiFe₂O₄-rGO through electrostatic gravitation force. Further, with decreasing pH from 9 to 3, increasing H⁺ ions in the solution, and increasing of electrostatic repulsion between positive surface of NiFe₂O₄/RGO and positive charge of BB41, adsorption efficiency is decreased (Fig. 10a–c).

Fig. 9.

3-D surface plot for the combined effect of pH and contact time (a) pH and adsorbent dose (b) pH and initial concentration of dye (c), contact time and adsorbent dose (d) contact time and initial concentration of dye (e) adsorbent dose and initial concentration of dye (f) on adsorption of DR81

Fig. 10.

3-D surface plot for the combined effect of pH and contact time (a) pH and adsorbent dose (b) pH and initial concentration of dye (c), contact time and adsorbent dose (d) contact time and initial concentration of dye (e) adsorbent dose and initial concentration of dye (f) on adsorption of BB41

Figures 9b, d, f and 10b, d, f show the increase of adsorption efficiency for the DR81 and BB41 with increasing of NiFe₂O₄/RGO dose at different concentrations and different contact time, this increase in adsorption is due to the increasing of available active sites on the NiFe₂O₄/RGO surface for dye molecules. Although, adsorption efficiency has been increased with increasing adsorbent dose, but the amount of adsorbed dye per unit mass of adsorbent is decreased due to the saturation of sites on the adsorbent. Similar trend has been reported for Modeling of arsenic removal from aqueous solution by means of MWCNT/alumina nanocomposite [50].

The effect of the contact time on the adsorption dye of DR81and BB41 is shown in Figs. 9a, d, e and 10a, d, e. These figures show that with increasing contact time the adsorption of dye on NiFe2O4/RGO increased, but this increasing trend is not considerable. Due to the large number of active sites adsorption on the NiFe₂O₄/RGO, the maximum adsorption rate occurs during the initial stages of the reaction up to first 5 min, and then with gradual occupancy of these sites the adsorption rate significantly becomes slow. After a lapse of time, the occupation of the remaining vacant sites adsorption on the adsorbent is more difficult for the dye molecules due to the repulsive force between the dye molecules and bulk solution [51].

The plot of the effect of dye concentration on adsorption efficiency is shown in Fig. 9c, f for DR81 and Fig. 10c, f for BB41. As can be observed with increasing dye concentration, the adsorption efficiency is decreased for both dyes. The reasons for the decreasing trend of the DR81dye adsorption by increasing concentration of dye may be explain by this point that in high concentration of dye, the ratio of the molecules of dye to the active sites of adsorption on NiFe₂O₄/RGO is high, and with the saturation of active sites on adsorbent, there will not be enough capacity for adsorption of further molecules and thus the chance of dye molecules are decreased for binding on the adsorbent surface. However, in high concentration of dye, despite the reduction in adsorption efficiency, adsorption capacity has increased. The similar trends have been reported for the adsorption of anionic dye Reactive Blue 19 onto Fe3O4 functionalized with hyperbranched polyethylenimine (Fe@HPEI) [49].

Optimization

The main purpose of the optimization process was to find a combination of variables levels in which the maximum adsorption efficiency of dyes of DR81 and BB41 is occurred by magnetic NiFe₂O₄/RGO nanocomposite. For this purpose, software Design-expert version 7.0.0 (Stat-Ease, trial version), and numerical technique, the best operating conditions accessible were obtained in the range of studied variables to achieve the specific point that maximizes the efficiency (Fig. 11). In numerical optimization approaches, response was set in maximize, adsorbent dose was set in minimize, contact time and initial dye concentration was set in the studied range. According to the obtained results, under optimum condition, the maximum adsorption for DR81 (anionic dye) and BB41 (cationic dye) were observed selectively at pH value of 3, 9 Contact time of 19.68, 18.16 min, adsorbent dose of 0.02 g and dye concentration of 40 mg/L. maximum efficiency by the model at this condition was found to be 96.41% for DR81and 97.87% for BB41.

Fig. 11.

Numerical optimization of four independent variables, contact time, adsorbent dosage and dye concentration for adsorption DR81 (a) and BB41 (b) onto NiFe₂O₄/RGO

Adsorption isotherms

The adsorption isotherm models were used to investigate the adsorption behavior of the DR81 and BB41 dyes onto the NiFe₂O₄/RGO. For this purpose the experimental data were fitted with two of the most commonly used isotherm models, including the Langmuir and Freundlich isotherms. Langmuir isotherm assumes that, only monolayer adsorption occurs on adsorbent surface with a fixed number of adsorption sites and adsorbed molecules do not interact with each other. Also the Freundlich isotherm assumes that adsorption is multilayer reversible and adsorption energy decreases exponentially with the completion of adsorption sites. The plots of Langmuir and Freundlich isotherms are shown in Fig. 12. As observed from this figure, the Langmuir model is more suitable in the description of the adsorption behavior of DR81 dye on the NiFe₂O₄/RGO with higher values of R² of 0.9759, while in the case of BB41 dye the adsorption can be best described by the Langmuir model with the higher value of R ² 0.9753.the various parameters of the isotherm models are shown in Table 5.

Fig. 12.

Langmuir and Freundlich isotherm plots for adsorption of DR81, BB41onto the NiFe2o4 / RGO

Table 5.

Langmuir and Freundlich isotherm constants for DR81, BB41

Dye	Langmuir isotherm coefficient			Freundlich isotherm coefficient
	qm (mg/g)	KL (L/mg)	R2	KF (mg/g)	1/n	R2
DR81	3762.935	1.116362	0.9759	23.5933	0.2929	0.9164
BB41	3449.23	0.96827	0.9538	22.0298	0.3548	0.9753

Conclusions

The present study shows that the magnetic NiFe₂O₄/RGO nanocomposite has much potential as a powerful adsorbent for the rapid adsorption of anionic (DR81) and cationic dyes (BB41) from aqueous solution. RSM based on CCD was applied in order to evaluate the effect of main and interactive of influential variables on the adsorption of DR81, BB41 including pH, contact time, adsorbent dose and dye concentration and also to determine the optimum conditions of adsorption. The result of analyses of variance (ANOVA) demonstrated that the quadratic models developed for adsorption of both dye on NiFe₂O₄/RGO was statistically significant. Furthermore, the factors of pH, dye concentration and adsorbent dose, have the greatest effect on adsorption, respectively, while contact time have the lowest effect on adsorption of both dye. The process was optimized and maximum adsorption efficiency at optimum conditions was found to be 96.41% by model for DR81 at value of pH 3, contact time of 19.68 min, adsorbent dose of 0.02 g, and dye concentration of 40 mg/L, also in the case of BB41, the optimum conditions were obtained 97.87% at value of pH 9 contact time of 18.16 min, adsorbent dose of 0.02 g, and dye concentration of 40 mg/L. also the adsorption behavior of the DR81 and BB41 onto NiFe2O4/RGO was best described by the Langmuir and Freundlich isotherm, respectively.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.