Introduction

Pollution from dye wastewater has drawn worldwide public concern due to its high biological toxicity, carcinogenicity, teratogenicity, and mutagenicity in living organisms, especially human beings (Belpaire et al. 2015, Forgacs et al. 2004). Various methods such as chemical oxidation (Lucas and Peres 2006, Pu et al. 2017a), photochemical catalysis (Pu et al. 2017b), membrane separation (Ma et al. 2017, Netpradit et al. 2003), electrochemical degradation (Riera-Torres et al. 2011, Szpyrkowicz et al. 2001), and biodegradation (Malachova et al. 2013), have been developed for the purification of dye wastewater. However, the application of these techniques has been limited due to their inevitable drawbacks, such as high energy consumption, secondary pollution, and high operating costs. Beyond the techniques listed above, absorption is considered to be one of the strategies with the most potential due to its advantages of easy operation, low energy consumption, few harmful byproducts, the possibility of adsorbent recycling, etc. (Ma et al. 2018). Adsorption approaches may be based on various adsorbents such as chitosan (Pu et al. 2017d), activated carbon (Greenwald et al. 2015), clay (Abidi et al. 2015), and bio-waste (Kyzas et al. 2013). Unfortunately, the practical application of adsorbents has been limited by their poor utilization rate, difficult recovery, and low adsorption capacity (Akbari et al. 2002).

Graphene oxide (GO) and graphene-based composite materials are attracting tremendous research interest as a novel adsorbent for environmental purification due to their enormous specific surface area (1000–1217 m2 g−1) (Tang et al. 2010), large number of active sites (Dreyer et al. 2010), and excellent dispersibility (Cheng et al. 2017). Its hydrophilic oxygen-containing functional groups (e.g., carboxyl, hydroxyl, and epoxy functional groups) (Dreyer et al. 2010, Huang et al. 2016) allow GO to combine easily with hydrophilic and hydrophobic compounds (e.g., heavy metals, dyes, and sediment), which gives GO great potential in wastewater treatment (Liu et al. 2012). GO has been proven to possess great potential as superior adsorbent for removing many kinds of organic and inorganic contaminants including heavy metal ions (Peng et al. 2017b), uranium (VI) (Zhang et al. 2016b), fluoride (Barathi et al. 2014), methylene blue (Zhang et al. 2016a), dibenzothiophene (Dizaji et al. 2016), and polyphenols (Peng et al. 2017a). Recently, magnetic graphene oxide has been widely used as a new magnetic solid-phase extraction adsorbent in environmental purification applications (Aliyari et al. 2016). In general, magnetic adsorbents are often composed of magnetic nanoparticles dispersed in a cross-linked polymer matrix, including natural and synthetic polymers (Yao et al. 2012). However, these methods suffer many drawbacks such as high cost, complicated operation, and easy clogging.

In this study, we primarily focused on developing a recyclable, multi-functional, and high-efficiency adsorbent for dye wastewater purification. We prepared a superparamagnetic graphene oxide/Fe3O4 hybrid nanocomposite (MGO) via an in situ co-precipitation strategy, which was successfully obtained through a facile and green in situ synthetic approach. The common dyes Congo red (CR, anionic dye) and methylene blue (MB, cationic dye) were used to evaluate its adsorption properties. Batch experiments under varied influential parameters, such as initial pH, dye concentration, temperature, and coexisting ions, were conducted to assess the properties of MGO. The synthesized MGO demonstrated good performance for dye removal and a wide range of abilities to adsorb many types of dyes, including both anionic and cationic dyes. The results demonstrate potential for the application of graphene oxide in wastewater treatment.

Materials and methods

Materials

The graphite powder was purchased from Aladdin Reagent Factory (Shanghai, China). Sodium nitrate (NaNO3, ≥ 99.9%); sulfuric acid (H2SO4, 98%); potassium permanganate (KMnO4, 65–68%); hydrogen peroxide (H2O2, 30 wt%); ammonia (NH4OH, 25–28%); ethanol (C2H5OH, ≥ 99.9%); MB; and CR were provided by Chengdu Kelong Chemical Reagent Company (Sichuan, China). FeCl2·4H2O and FeCl3·6H2O were purchased from Tianjin Zhiyuan Chemical Reagent Company (Tianjin, China). All chemical reagents were analytical grade and used without further purification. Deionized water used for all experiments was generated from Milli-Q water purification system (Ulupure Corporation, CA).

Experimental section

Preparation of graphene oxide

Graphene oxide was synthesized from graphite powder via a modified Hummers method (Stankovich et al. 2006). Typically, 2.0 g graphite powder and 1.2 g sodium nitrate were added into 60 mL concentrated sulfuric acid (98%) solution in a three-neck flask immersed in an ice bath while maintaining mechanical stirring. A total of 8.8 g potassium permanganate was slowly added into the flask and the temperature was cooled to 278.15 K for 12 h. Subsequently, 72 mL distilled water was added dropwise into the mixture at 293.15 K, and the mixture was stirred at 323.15 K overnight. After the color of the mixture turned bright yellow, 22 mL H2O2 solution was added gradually into the mixture for 2 h. The mixture was washed successively with 5% HCl solution and deionized water until the pH of the rinse water became neutral, after which it was freeze-dried (Scientz-50F, China) for more than 32 h. Finally, 0.3 g GO was dispersed in 100 mL water by an ultrasonic cell disruptor to obtain a GO dispersion (3 mg mL−1).

Preparation of magnetic graphene oxide nanocomposite

The magnetic graphene oxide nanocomposite (MGO) was synthesized by a chemical in situ co-precipitation method. Briefly, 1.3 mmol FeCl2·4H2O and 2.6 mmol FeCl3·6H2O were dissolved in 400 mL deionized water under an N2 atmosphere. An appropriate molar ratio of FeCl2 and FeCl3 was adopted (1:2) to meet the stoichiometric requirement. To prevent the aggregation of the GO sheets in a low pH (pH = 1.89) environment, 0.7 mL NH3·H2O was used to adjust the pH before adding 100 mL GO. Then, 40 mL NH3·H2O was continuously added with vigorous stirring for 3 h at 353.15 K. Finally, the black precipitate was washed several times with deionized water and ethanol to remove the residual alkali solution, which was heated at 333.15 K in vacuum conditions for 8 h.

Batch adsorption experiments

Two typical cationic and anionic dyes, MB and CR, were used as representative contaminants to evaluate the adsorption properties of MGO. All batch adsorption experiments were performed in a constant temperature shaker. Briefly, adsorption experiments were carried out by adding 25 mg MGO powder into 50-mL dye solution in a shaker (298.15 ± 0.5 K) for 24 h to reach adsorption equilibrium. The adsorbent was removed by magnetism and the remaining concentration of dye was measured by a visible spectrophotometer (V-1100D, Mita, Shanghai) at wavelengths of 664 nm for MB and 497 nm for CR. The adsorption capacity (qe) of MGO was calculated using Eq. (1) (Ren et al. 2013) as the following:

$$ {q}_e=\frac{\left({C}_0-{C}_t\right)V}{m}. $$
(1)

The removal rates (η) of the dyes were determined by using Eq. (2):

$$ \upeta =\frac{C_0-{C}_t}{C_0} $$
(2)

where C0 (mg L−1) is the initial concentration of the test solution, Ct is the concentration of the test solution (mg L−1) at time t (min), V (L) is the volume of the test solution, and m (g) is the weight of the adsorbent. All experiments were performed in three parallel groups for accuracy.

To assess the effects of initial pH, different temperatures, and initial concentration of dye on adsorption, consistent conditions were maintained except when changing the experimental variable. To investigate the effect of initial pH on adsorption, pH was adjusted by adding a small amount of HCl (1 mol L−1) and NaOH (1 mol L−1) while the solution volume and other conditions were constant. To assess the effects of different temperature, the reaction temperatures were adjusted by the water bath from 298.15 to 328.15 K ± 0.5 K. The effect of the initial dye concentration was measured by adsorbing different concentrations of dye (from 50 to 1500 mg L−1). The effect of coexisting ions on adsorption was studied by adding known amounts of Na+, Ca2+, K+, NO3−, Cl−, and HPO4−. The concentration of the residual dye was analyzed at established time intervals during the adsorption process.

Desorption studies

To evaluate the stability and reusability of the adsorbent, the adsorbent was collected from solution after reaching equilibrium and washed several times with ethanol, NaOH (1 mol L−1) and deionized water before being reused for the next adsorption. The above process was repeated five times.

Characterization

The crystal structure of MGO was characterized by XRD (Ultima IV, Japan) employing CuKα radiation (λ = 0.154 nm, 40 kV) in the range of 3° to 80°. FTIR (Nicolet-1170 SX, USA) was used to analyze the functional groups of MGO with the KBr wafer technique. The scanning wavelength range was from 400 to 4000 cm−1. The surface morphology was characterized by SEM (SIGMA300, Germany) at a voltage of 20 kV. The micromorphology was observed by TEM (JEM2100, Japan) operated at 200 kV. The magnetic properties were recorded at room temperature with a vibrating sample magnetometer (VSM) (730 T Lakeshoper, America). Samples were pretreated by heating under vacuum at 403.15 K for 10 h.

Results and discussion

Preparation and characterization of MGO

The MGO composite was synthesized through a chemical in situ co-precipitation approach (Fig. 1a). The color of the mixture rapidly changed from dark brown to black after NH3·H2O addition (Fig. 1b), as the in situ co-precipitation reaction between the Fe3+, Fe2+ mixture, and alkali solution formed magnetic Fe3O4 nanoparticles with the following reaction (Yang et al. 2012):

$$ {\mathrm{Fe}\mathrm{Cl}}_2+{2\mathrm{FeCl}}_3+{8\mathrm{NH}}_4\mathrm{OH}\to {\mathrm{Fe}}_3{\mathrm{O}}_4\ \left(\mathrm{on}\ \mathrm{GO}\right)+{8\mathrm{NH}}_4\mathrm{Cl}+{4\mathrm{H}}_2\mathrm{O} $$
(3)
Fig. 1
figure 1

a) Schematic illustration of MGO preparation via in situ co-precipitation. The graphene oxide was prepared by a modified Hummer’s method. b) Photographs of GO solution, combined Fe3+ and Fe2+ solution, MGO suspension, and the obtained MGO powder

Full size image

The color of mixture turned from brown to black, indicating the formation of Fe3O4. The resulting MGO powder was a black sand-like powder (Fig. 1b).

The crystalline phases of the pure Fe3O4, GO, and synthesized MGO were identified with XRD (Fig. 2a). The characteristic diffraction peaks at 2θ = 30.10°, 35.4°, 43.11°, 56.28°, and 62.74°, corresponded with the planes at (220), (311), (400), (511), and (440), respectively, which is in strong accordance with Fe3O4 nanoparticles (JCPDS No. 19-0629) (Wang et al. 2015). The GO displayed a broad diffraction peak ranging from 15° to 30°. The XRD pattern of MGO exhibited similar XRD patterns to Fe3O4 and the characteristic diffraction peaks matched well with those of Fe3O4. There was a weak diffraction peak of GO detected in the MGO, possibly due to the strong diffraction peaks of Fe3O4, which made the GO peak (25°) appear weak, indicating that the crystalline structure of GO was not changed substantially during synthesis. The EDX element distribution of MGO is shown in Fig. S1. The peaks of O, C, and Fe indicated that Fe compounds were successfully introduced into the GO surface. Based on XRD analysis, we can describe the composite formed by Fe3O4 and GO.

Fig. 2
figure 2

a) XRD patterns of GO, MGO, and Fe3O4. b) FTIR spectra of GO, MGO, and pure Fe3O4 NPs. c) Magnetic hysteresis loops of MGO. Conditions: initial MB/CR concentration 100 mg L−1, adsorbent dosage 0.5 g L−1, temperature 298 ± 0.5 K, stirring rate 180 rpm

Full size image

The FTIR spectra of GO, Fe3O4, and MGO are shown in Fig. 2b. The pure GO curve showed obvious characteristic absorption peaks at 3397, 1737, 1627, 1399, and 1079 cm−1. The adsorption bands at 3397 and 1627 cm−1 were attributed to −OH stretching and C=C stretching vibrations. The bands at 1737 and 1399 cm−1 were mainly ascribed to C=O and O–C=O stretching vibrations. The band at 1079 cm−1 was attributed to the stretching vibration of sp2 hybridized of C–O–C. A comparison of GO to MGO showed that the characteristic peaks at 1737 and 1079 cm−1 disappeared, and a peak at 574 cm−1 emerged which was assigned to the Fe–O stretching vibration. Data analyses indicated that the Fe3O4 nanoparticles were successfully embedded in the graphene oxide. Therefore, it was evident that some functional groups of GO were reduced via the in situ co-precipitation process (Wan and Wang 2017). Moreover, the dye adsorption by MGO could be attributed to hydrogen bonding interactions between O–C=O, –OH, and the dye’s characteristic groups. Carboxyl O–C=O groups were responsible for interactions between the D–SO3− (D = dye molecule) and the MGO.

The magnetic performance of the prepared MGO nanocomposite was determined by VSM. Typical magnetization hysteresis loops were recorded and are illustrated in Fig. 2c. The saturation magnetization was found to be 17.38 emu g −1. The magnetization curve of MGO was close to an “S” curve and the coercivity was negligible (nearly zero), indicating that no magnetization remained after removing the external magnetic field (Ren et al. 2013). Therefore, the superparamagnetic behavior of nanocomposite MGO was established. Moreover, the inset in Fig. 2c illustrates the color change in the dye solution before and after adsorption. The adsorbent dosage was 0.5 g L−1 and the initial dye concentration was 100 mg L−1. After adsorption reaction, the MGO was easily separated from aqueous solution within 40 s under a magnetic field, and the supernatant was colorless, as shown in Fig. 2c. Thus, the obtained MGO had highly efficient adsorption (removal rate can reach 99%) properties and could be readily magnetically separated.

Subsequently, the morphology of GO and MGO was studied by SEM and TEM. SEM images of GO (Fig. 3a) revealed that the GO had a smooth surface with many wrinkles, which led to its large specific surface area and abundant active sites. The prepared GO presented a typical multi-layered structure and the surface was relatively rough (Fig. 3b). Transmission electron microscopy (TEM) was performed to elucidate the morphology and structure of the MGO compound. It was clear that the elliptical and circular Fe3O4 nanoparticles were uniformly dispersed on the translucent graphene oxide substrate with an average diameter of ca. ~ 9.8 nm (Fig. 3d, inset), which indicated that the Fe3O4 nanoparticles were successfully loaded on the GO sheets.

Fig. 3
figure 3

SEM images of a) GO and b) MGO. TEM images of c) GO and d) MGO and size distribution of Fe3O4 nanoparticles formed inside the MGO

Full size image

Adsorption experiments

The initial pH was one of the most important factors in the adsorption process, influencing not only the surface charge of the adsorbent but also the functional groups on the GO and the chemical properties of the dye solution. The anionic dye CR and cationic dye MB were used to investigate the pH sensitivity of MGO, and the experimental pH values varied from 3 to 11. The results indicated that the adsorption of MGO on MB was insensitive to pH value changes. The equilibrium adsorption capacity of MB was higher than 180.2 mg g−1, and increased smoothly upon increasing the solution pH (Fig. 4a). There are two acid–base equilibriums of MB and GO, represented by GO-H → GO− + H+ and MBH2+ → MB+ + H+ (Zhang et al. 2011). The amount of hydrogen ions increased with decreasing pH values, thus competing more with the MB cations. Due to the abundance of H+, GO-H was the main form of GO− which decreased the amount of MB adsorbed. In contrast, there was more GO− at higher pH values, which may have enhanced the electrostatic attraction between MGO and MB, thus increasing the adsorption capacity for MB. However, it can be clearly seen from Fig. 4b that the adsorption curve of CR showed a different pattern from that of MB. The equilibrium adsorption capacity decreased sharply at low pH values (pH = 3), dropping from 189.4 to 34.1 mg g−1. The reason for this phenomenon might be because CR is a reactive indicator dye, and its structure changed with the change in pH value. At a low initial pH range (pH < 3.5), the −NH2 of CR would protonate to −NH3+ (Fig. S2), which decreased the electrostatic repulsive force between CR and MGO. The decrease of electrostatic repulsion in acidic condition was favorable to physical adsorption. However, according to the adsorption kinetics, it was found that the adsorption were mainly chemical adsorption, and these protonated functional groups would not be conducive to chemical adsorption. When the pH was lower than the acid dissociation constant of oxygen-containing groups, these groups became positively charged (C–OH2+, C–COOH2+), as a result of which the main interactions between CR and MGO were chemical bonds (Ahmad and Kumar 2010). The active sites at the surface of the adsorbent may determine the adsorption capacity of MGO.

Fig. 4
figure 4

Effect of initial pH on MB (a) and CR (b) adsorption. Conditions: initial MB/CR concentration 100 mg L−1, adsorbent dosage 0.5 g L−1, temperature 298 ± 0.5 K, and stirring rate 180 rpm. Effects of adsorption times and initial dye concentrations of MB (c) and CR (d) on MGO adsorption capacity. Effect of initial MB (e) and CR (f) dye concentration on adsorption of MGO, Fe3O4, and GO. Conditions: adsorbent dosage 0.5 g L−1; temperature 298 ± 0.5 K; stirring rate 180 rpm; initial concentrations: 50, 100, 200, 300, 500, 800, 1000, 1500 mg L−1

Full size image

With increasing pH, the zeta potential of MGO becomes negative. The zeta potential value decreased steadily from 12 to − 50 mV (Fig. S3) as the pH increased from 2 to 11. It was observed that the adsorption capacity of MB increased with decreasing zeta potential value, due to the increase of the Coulombic force between MB molecules and MGO.

The effect of initial concentration was systematically investigated. Figure 4c, d compare the adsorption performance at various initial concentrations (i.e., 50, 100, 200, 300, 500 and 800 mg L−1 of MB and CR). The results showed that the dye adsorption capacity gradually increased as the initial concentration increased. The adsorption capacity of MGO increased rapidly during the first 5 min, and the initial concentration of the solution had little effect on the adsorption rate (Fig. 4c, d). As the concentration of the dye solution increased, there was a significant increase in adsorption capacity. After a long period of time, the adsorption capacity increased slowly and eventually reached equilibrium. The surface sites may have been occupied by dye molecules, causing a decrease in adsorption rate.

Furthermore, Fig. 4e, f compare the different adsorption properties of MGO, Fe3O4, and GO for dye removal. The theoretical maximum equilibrium adsorption capacity of MGO, GO, and Fe3O4 were 1.4615, 2.8974, and 0.0591 mmol g−1 for MB and 0.9028, 0.7476, and 0.1721 mmol g−1 for CR (Table S3), respectively. The adsorption capacity of CR (36.2674 mg g−1) on Fe3O4 was better than that of MB (12.8027 mg g−1) because the Fe3O4 was positively charged in aqueous solution and therefore favored CR adsorption. The theoretical maximum adsorption capacities of MB (193.0502 mg g−1) and CR (150.3759 mg g−1) on MGO have similar values. However, upon comparing the dyes’ KL values and rates of adsorption (r), the MB adsorption rate (KL = 0.0885, r = 0.75) was much faster than that of CR (KL = 0.0178, r = 0.02) (Table S3). The adsorption of MB (225.225 mg g−1) on GO was significantly higher than that of CR (138.1215 mg g−1), which may be due to difference in the dyes’ structures and characteristics.

The BET surface areas of GO and MGO were measured as 1115.3468 and 407.1267 m2 g−1, respectively, suggesting that the introduction of Fe3O4 to GO decreased the surface area of the GO. However, MGO could be conveniently separated from the solution by magnetic field which increases its application prospects.

Adsorption kinetics

The adsorption kinetics reflect the state of the adsorption process throughout the whole adsorption reaction (Ho and McKay 1998). The mechanism of adsorption can be explored according to adsorption kinetic models. To reveal the most suitable mechanism for the adsorption process, pseudo-first- and pseudo-second-order models were used to analyze the adsorption data. The pseudo-first-order kinetic model is expressed as follows (Vadivelan and Kumar 2005):

$$ \mathit{\ln}\ \left({q}_e-{q}_t\right)={lnq}_e-{k}_1t $$
(4)

The pseudo-second-order kinetic model is represented by the following (Ho 2006, Ho and McKay 1999, Vadivelan and Kumar 2005):

$$ \frac{t}{q_t}=\frac{1}{{k_2{q}_e}^2}+\frac{t}{q_t} $$
(5)
$$ r={k}_2C $$
(6)

where qe and qt are the amounts of MB and CR adsorbed (mg g−1) at equilibrium and at time t (min), respectively; k1 is the pseudo-first-order rate constant (min −1); k2 is the pseudo-second-order rate constant (g mg−1 min −1); C is the concentration of the solution (mg L−1); and r is the rate of adsorption (g L−1 min−1).

The correlation coefficients and rate constants of the pseudo-first- and pseudo-second-order model are shown in Table S2, and Fig. S4. It can be seen that the R2 values of the pseudo-first-order model resulted in low correlation coefficients (0.5897 to 0.9835 for MB, 0.8988 to 0.9802 for CR). The R2 values of the pseudo-second-order kinetic correlation coefficient were close to 0.9999, which is higher than the pseudo-first order kinetic correlation coefficient (Table S2, Fig. S4). There were great discrepancies between the theoretical adsorption capacities and the experimental data. The maximum theoretical adsorption capacities (505.05 mg g−1 for MB, 492.61 mg g−1 for CR) of the pseudo-second-order kinetics model were consistent with the experimentally measured data. Therefore, it can be concluded that the adsorption of MB and CR by MGO fitted the pseudo-second-order kinetic model well, and the adsorption process may be chemical adsorption.

Furthermore, as the initial concentration increased from 50 to 500 mg L−1, the second-order rate constants (k2) of adsorption for MB and CR were found to decrease from 3.95 × 10−2 to 1.0 × 10−3 (g mg−1 min−1), and from 3.0 × 10−4 to 1.0 × 10−4 (g mg−1 min−1), respectively. The reaction rate was related to the solution concentration, which was consistent with previously reported studies (Demirbas et al. 2009, Hanafiah et al. 2012).

Adsorption isotherms

The two adsorption models were used to interpret the adsorption isotherm at different concentrations and temperature ranges. The Langmuir and Freundlich sorption models were utilized to match the adsorption isotherms. The Langmuir isotherm model assumes that all adsorption sites are subjected to the same monolayer surface adsorption and that the adsorbed molecules are completely independent. The Freundlich model applies to multi-layer adsorption, and the adsorption capacity of varies from site to site.

The equation of the Langmuir isotherm is given as follows:

$$ \frac{C_e}{q_e}=\frac{1}{k_L{q}_m}+\frac{C_e}{q_m} $$
(7)

where qe is the amount of dye adsorbed (mg g−1) at equilibrium, Ce is the equilibrium concentration of the solution (mg L−1), qm is the maximum adsorption capacity (mg g−1), and kL is the adsorption equilibrium constant (L mg−1), which is related to the affinity of the binding site.

The Freundlich isotherm equation is expressed as follows:

$$ \log\ {q}_e=\log\ {k}_F+\frac{1}{n}\log {C}_e $$
(8)

where kF is the Freundlich constant, and n is the heterogeneity factor.

The theoretical correlation coefficient (R2) parameters of the adsorption isotherms are presented in Table S3 and Fig. S5. The R2 values of the Langmuir and Freundlich models are close to unity: 0.997 and 0.90 for MB and 0.995 and 0.94 for CR. These values imply that the Langmuir model may better describe the adsorption isotherm for the adsorption of these two different dyes. The correlation coefficient R2 for the Langmuir equation is higher than the linear correlation coefficients for the Freundlich model. Furthermore, the theoretical maximum equilibrium adsorption capacity of the Langmuir equation for MB and CR was 613.50 and 909.09 mg g−1 at 328.15 K, respectively, which was closer to the experimental data. It was concluded that the adsorption of MB and CR by MGO was better described with the Langmuir model. The adsorption process was mainly monolayer uniform adsorption and mainly occurred on the functional groups of adsorption sites on the surface of MGO, such as hydroxyl, carboxyl, and epoxy functional groups. The monolayer coverage of dye molecules on the surface of the adsorbent may be explained at the molecular level. The structure of graphene oxide is layered, and there was no evidence of substitution taking place between different layers.

The maximum adsorption capacity of MGO increased from 546.45 mg g−1 (at 298.15 K) to 613.50 mg g−1 (at 328.15 K) for MB and from 628.93 mg g−1 (at 298.15 K) to 909.09 mg g−1 (at 328.15 K) for CR, indicating that the adsorption capacity increased with increasing temperature (Table S4). This trend could be due to the enhanced mobility of the MB and CR molecules with rising temperature (Dawood 2012). The dye molecules may obtain adequate energy to interact with active sites at the surface of the adsorbent. The adsorption equilibrium constant kL increased with increasing temperature and with greater adsorption capacity. This can be explained by the Brownian motion, which causes liquid molecules to move more intensely at higher temperature. Thus, the impact forces on dye particles by liquid molecules from different directions increase with rising temperature, resulting in increased adsorption capacity.

The suitability of the predicted adsorption model was informed by the value of the separation factor constant (RL) which indicates the probability of the adsorption process. The separation factor (RL) can be inferred from the Langmuir equation and indicates whether the adsorption process is favorable. The values of this parameter mean that the adsorption isotherm is unfavorable (RL > 1), linear (RL = 1), favorable (0 < RL < 1), or irreversible (RL = 0) (Chang et al. 2007, Vimonses et al. 2009). The separation factor (RL) is defined as follows:

$$ {R}_L=\frac{1}{1+{k}_L{C}_0} $$
(9)

where kL (L mg−1) is the Langmuir constant, and C0 (mg L −1) is the initial dye concentration. In this study, the values of RL were within the range of 0–1, indicating that the adsorption of MB, and CR onto MGO was a favorable process.

As an interesting material with enormous specific surface area, GO has abundant active sites that can effectively interact with dyes. The possible mechanism can be explained by electrostatic interact, hydrogen bonds, Van der Waals force and hydrophobic–hydrophobic interactions; a more in-depth explanation is as follows: (i) hydrogen bonds occur between the dye molecule and oxygen-containing hydroxyl (–OH) and carboxyl (–COOH) groups. (ii) Ion exchange occurs between the oxygen-containing groups (–OH, –COOH) in MGO and D-SO3− (D-SO3−, D = dye molecule)), forming MGO–COO–SO2–D and MGO–C–O–SO2–D (Guo et al. 2016). Guo et al. (2016) have reported that the –SO3− groups of CR are involved in the adsorption process. (iii) The protonation of –COOH and –OH occurred on the surface of MGO in acidic solution (i.e., MGO–OH + H+ ↔ MGO–OH2+) (Ahmad and Kumar 2010).

Thermodynamic study

To evaluate the effect of temperature on adsorption process, the thermodynamic parameters Gibbs free energy (∆G0, kJ mol−1), enthalpy (∆H0, kJ mol−1), and entropy (∆S0, kJ mol−1 K−1) can be calculated by the following equations:

$$ \ln \left({k}_L\right)=\frac{\Delta {S}^0}{R}-\frac{{\Delta H}^0}{RT} $$
(10)
$$ {k}_L=\frac{q_e}{c_e} $$
(11)
$$ \Delta {G}^0={\Delta H}^0-T\Delta {S}^0 $$
(12)

where kL is the Langmuir constant (L g−1), R is the universal gas constant (8.314 J mol−1 K−1), and T is the absolute temperature (K).

Table 1 shows the thermodynamic parameters of the adsorption. The negative values of ∆G0 indicated that the adsorption reaction can be carried out at room temperature and the decrease of ∆G0 with increasing temperature indicated that the adsorption reaction was spontaneous at higher temperatures. The values of ∆H0 and ∆S0 were found to be 113.14 and 474.16 J K−1 mol−1 for MB and 41.83 and 213.65 J K−1 mol−1 for CR, respectively. The positive values of ∆H0 and ∆S0 showed that the adsorption of MB and CR by MGO is endothermic. Therefore, the prepared MGO could be effectively used in dye adsorption without harsh reaction conditions, indicating that the adsorbent has the potential possibility to be used in practical wastewater treatment.

Table 1 Thermodynamic parameters for MB and CR adsorption onto MGO
Full size table

Effect of coexisting ions

In practical applications, the large amount of dissolved metal cations and inorganic anions, such as Na+, Ca2+, K+, NO3−, Cl−, and HPO4−, may greatly affect the adsorption process. Therefore, the effect of individual coexisting ions (1 mol L−1) on the adsorption of MB and CR was investigated in series. Figure 5 summarizes the effect of the common coexisting metal cations Na+, Ca2+, and K+ and inorganic anions NO3−, Cl−, and HPO4−, all of which showed no significant effects on MB/CR adsorption. Hence, MGO composites could serve as an effective adsorbent for dye wastewater remediation.

Fig. 5
figure 5

Effect of coexisting ions. Conditions: adsorbent dosage 0.5 g L−1; temperature 298 ± 0.5 K; stirring rate 180 rpm; initial dye concentrations 100 mg L−1; Na+, Ca2+, K+, and NO3-, Cl−, and HPO4- concentrations 1 mol L−1

Full size image

Desorption and regeneration

The continuous recycle experiments were conducted to evaluate the stability and the application potential of the prepared adsorbent. The obtained MGO particles were added to the MB/CR solution (50 mg L−1), and the residual concentration was measured after reaching adsorption equilibrium. Distilled water and NaOH were used to desorb dye several times until no more leaching dye could be detected. The adsorption rates of MB and CR still were maintained at 97.43% and 87.23% after five successive adsorption-desorption cycles (Fig. 6). These results indicated that the performance of MGO was not appreciably impaired after five cycles, which illustrated its good stability and practical application potential.

Fig. 6
figure 6

Regeneration and reuse of MGO. Conditions: adsorbent dosage 0.5 g L−1; temperature 298 ± 0.5 K; stirring rate 180 rpm; initial concentrations 50 mg L−1

Full size image

Table 2 compares the maximum adsorption capacity of graphene-based adsorbents for dye removal under different experimental conditions. The maximum adsorption capacities of dyes on MGO in this study were better than those of most previously studied graphene-based materials. MGO has a slight advantage in adsorption capacity. Since it could also be efficiently recycled and reused, MGO has broad practical application prospects in the treatment of organic pollutants.

Table 2 Comparison of MGO with other graphene-based adsorbents
Full size table

Conclusions

In summary, a feasible, economical and green route for the removal of two typical dyes by a magnetic graphene oxide/Fe3O4 (MGO) nanocomposite was demonstrated. The results showed that the MGO exhibited a great adsorption ability for dye removal and could adapt to a wide pH range from 5 to 11. The adsorption behavior was best suited for the Langmuir adsorption isotherm equation, showing calculated maximum adsorption capacity of 546.45 and 628.93 mg g−1 for MB and CR, respectively. The adsorption process fit the pseudo-second-order kinetic model well, which revealed that it might be the chemical interaction between adsorbate and adsorbent. Furthermore, the prepared MGO nanoparticles maintained good stability after five successive adsorption-desorption cycles. The results show that this “green” magnetic graphene oxide adsorbent has a good stability and high adsorption capacity and may serve as a promising absorbent for wastewater purification.