Graphene oxide-based polymeric membranes for broad water pollutant removal

Abstract

Graphene oxide (GO) and its derivatives display excellent removal abilities of water contaminants; however, the complex preparation process of GO-based adsorbents and difficult collection of GO sheets during the adsorption process limit their practical applications. Hence, three kinds of GO-based polymeric membranes with specific adsorption characteristics were fabricated by a facile blending method, including GO/PES membrane, reduced GO (RGO)/PES membrane, and polyethyleneimine (PEI) coated GO membrane of GO@PEI/PES membrane. The GO/PES membrane exhibited selective adsorption for cationic dyes, the RGO/PES membrane exhibited selective adsorption for endocrine disruptors, and the GO@PEI/PES membrane exhibited selective adsorption for anionic dyes. The adsorption data fitted the pseudo-second-order kinetic model and the Langmuir isotherm well, and the adsorption process was controlled by the interparticle diffusion. The thermodynamic studies indicated that the adsorption reactions were spontaneous and exothermic processes. The dynamic adsorption results indicated that the prepared membranes could be used in wastewater filtration. The study indicated that GO-based polymeric membranes with broad water pollutant removal could be fabricated by facile strategies, and the problem of difficult collection of GO sheets during and after adsorption process was solved.

---

1. Introduction

Millions of people worldwide are suffering from the shortage of fresh and clean drinking water. Rapid developments of industrialization, population expansion, and unplanned urbanization have largely induced the severe water pollution and surrounding soils. The main sources of freshwater contamination are caused by dumping of industrial effluent, runoff from agricultural fields and discharge of untreated sanitary wastewater.¹ The unbridled emission has led to the existence of various kinds of toxic and harmful substances in water, such as dyes and endocrine disrupters, that danger both human life and ecological environment.² Dyes are found in the wastewater streams of industrial processes, including paint manufacture, dyeing, textiles, paper, and others. Dyes can present toxic effects to ecological life, reduce light penetration and color water even at ultralow concentrations.^3,4 Most endocrine disrupters are used as raw materials to manufacture chemical products, pharmaceuticals and personal care products.^5,6 They are released into water during manufacturing processes and/or leaching from final products. Endocrine disrupters can interfere with the normal functioning of the endocrine system which will affect health, reproduction, and development of humans and wildlife.^7,8 As a result, water purification is becoming a global concern and various methods have been employed for the removal of hazardous substance from wastewater.⁹ Among the methods, adsorption is the most attractive and effective technique due to its high efficiency, low-cost and easy design.³ Till now, it has been reported that activated carbon,^1,3 porous polymers,^10,11 natural minerals^12–14 and macromolecules^15,16 could selectively and efficiently remove water pollutants.

Graphene oxide (GO) is one of the important derivatives of graphene and can be considered as a precursor for graphene synthesis by either chemical or thermal reduction processes.¹⁷ It has a layered structure with oxygen functional groups bearing on the basal planes and edges,¹⁸ which endows it an ultra-large specific surface area and negatively charged surface. In addition, the incomplete oxidation of GO makes it retaining the structures of aromatic rings and carbon–carbon double bonds.¹⁹ As a result, GO sheets exhibit outstanding adsorption capacity for cations, aromatic hydrocarbons and metal ions^20–22 through electrostatic interaction, π–π bonding interaction and Lewis acid–base interaction.²³ Meanwhile, the functional groups of GO such as hydroxyl, carboxy, epoxy groups and double bonds can serve as the sites for chemical modification or functionalization by many well-known chemistry strategies. Therefore, many chemical species can be immobilized on the surface of GO through either covalent or noncovalent bonds for the design of various features of GO sheets,¹⁷ thus an increasing number of GO-based water scavengers with ingenious functions are emerging constantly.

However, since GO belongs to nano-materials and presents high dispersibility in aqueous solution,¹⁷ the practical application and collection of GO sheets become difficult during and after the adsorption process, and the residual GO may lead to potential nano-toxicity to aqueous creatures. These problems restrict the practical applications of GO as adsorbents on a large scale. To solve the problems, various methods have been applied. Magnetic nanoparticles were deposited onto the GO sheets, and the adsorbents could be controlled and separated by a magnetic field during the adsorption process;^24,25 three-dimensional GO-based macro-materials, such as GO-based aerogels^26,27 and hydrogels^28,29 could also be used as adsorbents; in addition, GO could be dispersed well in some polymer/solvent systems to prepare the composites with particular morphology, such as particles,³⁰ fibers³¹ and membranes.³² Compare with powders, particles and gels, polymeric membranes have been more widely adopted in water treatment due to its porous structure, good mechanical property and relatively low cost.³³ The functionalization of polymeric membranes could be designed via grafting, coating, blending and so on. Among them, blending is simple to be applied in industrial fields. Therefore, we anticipate that GO-based polymeric membranes could be facilely designed via blending to reach versatile applications in wastewater treatment.

In this study, three kinds of GO-based polymeric membranes with different adsorption performances were fabricated by a facile blending method followed with a liquid–liquid phase inversion technique: (i) GO/PES membrane, GO was dispersed in polyethersulfone (PES) solution and then to prepare membrane; (ii) RGO/PES membrane, reducing GO/PES membrane directly with hydrazine hydrate solution; and (iii) GO@PEI/PES membrane, GO was coated by polyethyleneimine (PEI), and then blended with PES solution to prepare membrane. PES was chosen as the polymeric matrix because it showed outstanding oxidative, thermal, and hydrolytic stabilities, as well as good mechanical properties.³⁴ Cationic dye methylene blue (MB), anionic dye amaranth (AR) and endocrine disruptor bisphenol A (BPA) were chosen as templates to verify the specific adsorption for the three membranes. In addition, the adsorption kinetics, thermodynamics, isotherms and dynamic adsorptions of each membrane for different adsorbates were investigated and discussed.

2. Materials and methods

2.1. Materials

Graphite flakes were obtained from Sigma-Aldrich. Polyethersulfone (PES, Ultrason E6020P) was purchased from BASF chemical Co. (Germany) and was dried at 60 °C for 12 hours before use. Polyethyleneimine (PEI, MW = 70 000, 50% aqueous solution), methylene blue (MB), amaranth (AR) and bisphenol A (BPA) were obtained from Aladin reagent Co. Ltd. (China). All the other reagents were obtained from Changzheng chemical reagent Co. (Chengdu, China) and used without further purification. Dialysis membranes (MWCO = 8000–14 000 Da) were obtained from Solardio (Canada). Deionized water was used throughout the studies.

2.2. Preparation of graphene oxide (GO) and PEI coated GO (GO@PEI)

GO was prepared by a modified Hummers method, the detail was shown in the ESI.† The preparation of GO@PEI is as follows: 100 mL of diluted GO aqueous solution (1 mg mL⁻¹) was poured into a flask, and then certain amount of PEI was added until its concentration reached 15 mg mL⁻¹. The mixture was vigorously stirred by a magnetic stirrer and sonicated in a 40 kHz sonic bath until it became homogeneous solution. Then the product was stirred overnight, the excess PEI was removed and the prepared GO@PEI was concentrated by centrifugation.

2.3. Preparation of GO-based polymeric membranes

First of all, the water in GO solution was replaced with N,N-dimethyl acetamide (DMAC) and the GO/DMAC solution was concentrated to a required concentration by centrifugation. Then PES was added and the mixture was stirred by a magnetic stirrer for 12 hours to get homogeneous solution. Before fabricating the membranes, the solution was treated using sonication for 10 minutes and degassed by a vacuum pump. The GO/PES membrane was prepared by casting the solution on a clean glass plate at room temperature using a knife with a thickness of 100 μm. The glass plate with the casting solution was immersed in a water coagulation bath, and then the membrane was formed immediately due to liquid–liquid phase inversion. Subsequently the prepared membrane was dipped in fresh DI water to remove residual DMAC. For the sake of comparison, the membranes with two different concentrations of GO were prepared, i.e., GO/PES-3 and GO/PES-6, and the concentrations of GO in DMAC were 3 and 6 mg mL⁻¹, respectively.

For the fabrication of RGO/PES-3 and RGO/PES-6 membranes, GO/PES-3 and GO/PES-6 membranes were placed in the bottom of a beaker with a strainer, respectively. Then 250 mL of DI water and 5 mL of hydrazine hydrate (80 wt%) was added. The solution was stirred by a mechanical stirring at 80 °C for 12 hours. Subsequently the prepared membranes were dipped in fresh DI water to remove residual chemicals.

GO@PEI/PES-3 and GO@PEI/PES-6 membranes were prepared with the same method of GO/PES membranes by replacing GO with GO@PEI. Pure PES membrane was also prepared as a reference. The concentration of PES of all the samples was 16 wt%.

For the sake of characterization, reduced GO (RGO) was obtained by re-dissolving the RGO/PES membrane in DMAC, the dissolved PES was removed by centrifugation, and the sediment RGO was washed by DMAC and DI water and dried at 60 °C.

In the following discussion, pure PES, GO/PES-3, GO/PES-6, GO@PEI/PES-3, GO@PEI/PES-6, RGO/PES-3 and RGO/PES-6 membranes were simply abbreviated to M-PES, M-GO3, M-GO6, M-PEI3, M-PEI6, M-RGO3 and M-RGO6, respectively.

2.4. Characterization of GO-based materials

The prepared GO, RGO and GO@PEI samples were characterized using atomic force microscopy (AFM), Fourier transform infrared spectroscopy analysis (FTIR) and thermal gravimetric analysis (TGA). AFM images were taken on a Multi-Mode Nanoscope V scanning probe microscopy system (Vecco Instruments Co., USA). The samples for AFM were prepared by dropping aqueous GO-based materials (∼0.01 mg mL⁻¹) on freshly cleaved mica surface and dried under vacuum at 60 °C. FT-IR patterns were recorded on Nicolet 560 (Nicolet Co., America) instrument. Aqueous GO-based materials were firstly dried in an oven at 60 °C and ground with KBr together, and then pressed into a pellet for the FTIR characterization. Thermo-gravimetric analysis (TGA) was performed by using a TG209F1 (Netzsch Co., Germany), and the dried samples were heated at a rate of 15 °C min⁻¹ from ambient temperature to 600 °C under N₂ atmosphere.

2.5. Characterization of membranes

The morphology of GO-based hybrid PES membranes were observed by scanning electron microscopy (SEM); meanwhile, the pore size distribution, surface zeta-potential, water content and hydrodynamic permeability were also measured.

For the SEM (JSM-7500F, JEOL) observation, the membranes were freeze-dried, cut by a single-edged razor blade after immersing into liquid nitrogen, and then attached to the sample holder, coated with a gold layer.

For the surface zeta-potential measurement, the dried membranes with the area of 1 × 3 cm² was placed in a flat surface cell and measured by a Delsa™ Nano C Particle Analyzer (Beckman Coulter, American) at room temperature.

The water content of membranes (W) was determined from the weight change between the wet and the dry membranes in the unit gram of dry membrane based on eqn (1):

where M_w and M_d are the weights of the wet and dry membrane samples, respectively.

In order to measure the hydrodynamic permeability, a dead-end ultrafiltration cell with an effective area of 3.9 cm² was used. The membrane was firstly pre-compacted by DI water for 30 minutes to get steady filtration, then the water flux (F_w) was measured, and expressed as the hydrodynamic permeability using eqn (2):

where V is the volume of DI water (mL); S is the effective membrane area (m²); P is the pressure applied to the membrane (mmHg); t is the time (h).

2.6. Batch adsorption studies

Methylene blue (MB), bisphenol A (BPA) and amaranth (AR) were dissolved in DI water as the stock solutions (2 mmol L⁻¹; BPA was dissolved in a small amount of ethanol firstly) and further diluted with DI water to required concentrations before use. All adsorption experiments were performed in sealed 20 mL glass bottles that contained 5 pieces of membranes with the area of 1 cm² and 10 mL of MB, BPA or AR solutions in the appropriate concentrations. The bottles were placed in a water bath at a shaking speed of 100 rpm at certain temperatures.

In order to investigate the adsorptive selectivity of the prepared membranes, M-PES, M-GO3, M-GO6, M-RGO3, M-RGO6, M-PEI3 and M-PEI6 were applied in 100 μmol L⁻¹ MB, BPA and AR solutions separately at 25 °C. After 12 hours, the concentrations of the solutions were determined with the UV-vis spectrophotometer 756PC at the wavelength of 631 nm for MB, 280 nm for BPA or 521 nm for AR, respectively. The removal ratio (R) was determined using eqn (3):

where C₀ is the initial concentration of the solution (μmol L⁻¹); C_t is the concentration at the time t (μmol L⁻¹).

For the adsorption kinetic study, the prepared membranes were applied in 100 μmol L⁻¹ MB, BPA and AR solutions at 25 °C, respectively. The concentration of each solution was determined at different time intervals from 0.5 to 12 hours.

To study the effect of the initial concentrations on the adsorption behavior, the prepared membranes were applied in MB, BPA and AR solutions with defined concentrations (25 to 300 μmol L⁻¹) at 25 °C, respectively. The concentration of each solution was determined after 12 hours, and the data were also used for analyzing adsorption isotherm.

To investigate the adsorption thermodynamics and the effect of temperature on the adsorption behavior, the prepared membranes were applied in 100 μmol L⁻¹ MB, BPA and AR solutions separately in certain temperatures (20 to 60 °C). The concentration of each solution was determined after 12 hours.

2.7. Dynamic adsorption studies

The method of dynamic adsorption studies is as follows: the prepared membrane with an effective area of 3.9 cm² was fixed in a dead-end ultrafiltration cell. MB, BPA or AR solutions with the certain concentrations were separately pumped through the cell with certain flow rate by using a peristaltic pump. The solutions were collected at prescribed time intervals and the concentrations were measured.

3. Results and discussion

3.1. Preparation and characterization of GO-based materials and membranes

The GO aqueous solution was prepared by the modified Hummer's method. PEI coated GO sheets (GO@PEI) were prepared through electrostatic attraction. As illustrated in Fig. 1a, adding PEI in GO aqueous solution, the positively charged PEI would coat onto the negatively charged GO sheets spontaneously. GO/PES membranes and GO@PEI/PES membranes were prepared by a liquid–liquid phase inversion technique.¹¹ The reduced GO (RGO)/PES membranes were fabricated by dipping GO/PES membranes in N₂H₂ solution directly, instead of preparing RGO solution followed by blending with PES; since the reducing process would usually influence the dispersibility of GO in water, the accumulation and agglomeration would occur among RGO sheets.¹⁹ By reducing GO/PES membrane directly, the agglomeration of RGO could be avoided and the homogeneous RGO/PES membrane could be obtained simply.

Fig. 1 The schematic illustration of membrane fabrication (a); the digital photographs of the prepared membranes (b).

The digital photos of the prepared membranes are shown in Fig. 1b. All membranes displayed uniform colors; by reason of blending with GO, the colors of the membranes went from white (M-PES) to brown and gray (M-GO3 and M-GO6), and the color changing became more obvious when the content of GO increased. After the reduction by N₂H₂ solution, M-GO3 and M-GO6 changed into black (M-RGO3 and M-RGO6). M-PEI3 and M-PEI6 also showed brown and gray colors but darker than M-GO3 and M-GO6, that could be ascribed to the slight reduction of GO during the introduction of PEI.²⁸

3.1.1. Characterization of GO-based materials. The surface morphology and the height of GO-based materials were characterized by AFM. As shown in Fig. 2a, GO exhibited flat sheet with an average thickness of ∼1.0 nm, verifying the characteristic single-layered 2D structure. The RGO sheet was similar to GO, as shown in Fig. 2b. Since the surface groups were unlikely to be reduced entirely,³⁵ the height of RGO slightly decreased, and the thickness of RGO sheet was about 0.8 nm. As shown in Fig. 2c, after coating with PEI, the GO@PEI sheet also exhibited homogeneous surface; and the thickness increased to ∼3.2 nm. It confirmed that PEI has been successfully immobilized onto the GO sheet.

Fig. 2 Characterization of the GO, RGO and GO@PEI sheets: AFM image of GO (a), RGO (b) and GO@PEI (c); FTIR spectra (d); and TG curves (e).

The chemical compositions of GO-based materials were confirmed by FTIR patterns. As shown in Fig. 2d, multiple peaks for GO were observed in the range of 900 to 1500 cm⁻¹, which could be assigned to the functional groups, such as C–O–C and C–OH. The band at 1723 cm⁻¹ is associated with the ν(C=O) in carboxylic acid and carbonyl moieties;³⁶ in the RGO pattern, the peak at 1723 cm⁻¹ disappeared, verifying the decrease of C=O in carboxylic acid and carbonyl moieties because of the reduction of GO sheets; compared with GO pattern, the increase in two peaks at 2921 and 2831 cm⁻¹ could be observed in the GO@PEI pattern, which could be attributed to symmetric and asymmetric stretching modes of H–C–H of the PEI chains; meanwhile, two new bands at 1452 (C–N stretching vibration) and 1585 cm⁻¹ (N–H bending vibration) appeared in this pattern too, those results reflected the successful introduction of PEI.^28,37

The thermal stability of the prepared GO-based materials was characterized by TGA. Fig. 2e shows the weight loss of the samples as a function of temperature under nitrogen atmosphere. GO was thermally unstable and its mass loss at 240 °C was ∼35%, which could be ascribed to the decomposition of labile oxygen-containing functional groups.³⁸ Because of the decrease of the oxygenated groups after reduction, RGO exhibited less weight loss (∼10%) until the temperature rose to 600 °C, indicating the successful reduction of GO sheets. Compared with GO, GO@PEI showed a less mass loss around 240 °C, indicating the slight reduction of GO during the introduction of PEI.²⁸ The weight loss of GO@PEI between 240 and 600 °C could primarily be resulted from the thermal decomposition of PEI and the oxygen-containing functional groups on GO sheets.

3.1.2. Characterization of membranes. The morphologies of the surface and cross-section for the membranes were observed by SEM. The SEM pictures of M-PES, M-GO6, M-RGO6 and M-PEI6 are shown in Fig. 3. M-GO3, M-RGO3 and M-PEI3 had the same structures with M-GO6, M-RGO6 and M-PEI6, respectively, so the SEM images of them were not given. The images of each membrane were taken in three forms: the membrane surface with a magnification of 500×; the cross-section with a magnification of 1000×; and the cross-section with a magnification of 15 000×. As seen in Fig. 3, either blending with GO (M-GO6) or the subsequent reduction in N₂H₂ solution (M-RGO6) did not influence the surface morphology. However, an uneven surface with several holes could be observed after blending with GO@PEI (M-PEI6). From the cross-sectional images with the magnification of 1000×, it was found that the regularity of the pores with the finger-like structures in PES were interrupted by the introduction of GO sheets, irregular porous structures with smooth internal surfaces could be observed in M-GO6. After reduction, the internal surfaces in M-RGO6 became porous. The cross-sectional morphology of M-PEI6 was different from other membranes, and the messy and irregular pores appeared all over the entire cross-section. With the magnification of 15 000×, the gauze-like GO-based sheets could be observed in M-GO6, M-RGO6 and M-PEI6 (as indicated by red arrows), which were deposited and covered in the internal surfaces of the membranes. Water content could reflect the porosity of the membranes roughly, and the data are summarized in Table 1. All the samples exhibited higher water contents, indicating the porous nature of the membranes. Compared with PES, the water contents of GO-based membranes were increased obviously. Meanwhile, with the increase of GO contents, the water contents of both GO/PES and RGO/PES membrane increased; however, this phenomenon could not be observed in GO@PEI/PES membrane.

Fig. 3 The SEM images of M-PES, M-GO6, M-RGO6 and M-PEI6.

Table 1 Water contents, flux and zeta potentials of prepared membranes

	Water contents (%)	Flux (mL m^−2 mmHg^−1 h^−1)	Zeta potential (mV)
M-PES	245.7	270.1	−5.17
M-GO3	284.2	1493.7	−11.59
M-GO6	334.1	2249.2	−17.33
M-RGO3	284.3	1219.8	−7.31
M-RGO6	318.9	1874.3	−14.23
M-PEI3	289.3	1731.6	18.56
M-PEI6	281.7	1783.2	23.29

---

The pore structure of PES membrane would be affected by adding other materials during the formation process,^8,39 and the porosity would usually increase with adding hydrophilic materials.⁴⁰ So the higher porosity could be obtained in GO/PES and RGO/PES membranes due to the hydrophilic nature of GO sheers. Because of the excellent hydrophilicity of PEI, the GO@PEI sheets became more hydrophilic than GO sheets; it might be the reason why the GO@PEI/PES membranes exhibited such a special morphology. The similar phenomenon could also been found in our previous research.⁴¹

The pure water fluxes and zeta potentials of the membranes were measured and the data are also presented in Table 1. After blending with GO-based materials, the fluxes of the prepared membranes increased significantly. For GO/PES membranes, the flux increased with the increase of GO contents; the fluxes of the RGO/PES and GO@PEI/PES membranes exhibited the same tendency. The M-PES showed the zeta potential of about −5 mV, indicating the charge of PES was close to electric neutrality. Based on the negative charge nature of GO sheets, the zeta potentials of M-GO3 (−11 mV) and M-GO6 (−17 mV) decreased, and the value was more negative when the content of GO increased. As part of surface groups were not reduced by H₂N₂, the zeta potentials of M-RGO3 and M-RGO6 were still negative, about −7 mV for M-RGO3 and −14 mV for M-RGO6. Because of the positive charge nature of PEI, the M-PEI3 and M-PEI6 exhibited the zeta potentials of 18 and 23 mV, respectively.

3.2. The adsorption capacities and selectivity of membranes

A number of studies indicated that GO and GO-based adsorbents could adsorb cationic dyes effectively through electrostatic interaction.^26,42,43 In this study, the GO/PES composite membranes with good adsorption efficiency for cationic dyes were obtained by blending GO with PES (as shown in Fig. 1a). In the meantime, the membranes with adsorption capacity for endocrine disrupters were prepared after the reduction with GO/PES membranes. At last, through coating GO with PEI and then blending with PES, the membranes with increasing adsorption efficiency for anionic dyes were prepared. The removal ratios of each membrane for methylene blue (MB, cationic dye), bisphenol A (BPA, endocrine disrupter) and amaranth (AR, anionic dye) are shown in Fig. 4. The M-PES exhibited weak adsorption capacity for MB, BPA and AR; the M-GO3 and M-GO6 showed increased adsorption capacity for MB; the M-RGO3 and M-RGO6 showed increased adsorption capacity for BPA; and the M-PEI3 and M-PEI6 showed increased adsorption capacity for AR. Meanwhile, the M-RGO3 and M-RGO6 also showed good adsorption capacity for MB. The results of MB and AR removal could be observed intuitively, as shown in Fig. 4d. The adsorption capacity of GO/PES membranes for MB and GO@PEI/PES membranes for AR could be attributed to the electrostatic interaction. For the BPA removal capacity of RGO/PES membranes, two kinds of interactions might be considered: the one was the π–π interaction between the benzene rings of BPA and the RGO planes; the other was the hydrogen bonding between the oxygen-containing groups contained in both BPA and RGO.⁴⁴ As the oxygen-containing groups could not be reduced totally, the RGO/PES membranes were still negatively charged, so they had certain adsorption capacity for cationic dye.

Fig. 4 The removal ratios of the prepared membranes for MB (a), RGO (b) and AR (c) (the insets are the molecular formulas of MB, AR and BPA, respectively); the digital photographs of the adsorption results of the prepared membranes for MB (blue solutions) and AR (red solutions) (d).

3.3. Adsorption kinetics study

In order to better understand the mechanism of the adsorption process, adsorption kinetics was investigated. The effect of contacting time in the adsorption process is shown in Fig. 5a. The adsorption capacity of each membrane for their corresponding contaminant (GO/PES membranes for MB, RGO/PES membranes for BPA, and GO@PEI/PES membranes for AR) increased quickly in the first 4 hours and then rose slowly until the adsorption equilibrium was reached in 8 hours. The equilibrium adsorption capacities of M-GO3 and M-GO6 for MB were 27.95 and 35.31 mg g⁻¹, respectively; those of M-RGO3 and M-RGO6 for BPA were 12.44 and 16.85 mg g⁻¹; and those of M-PEI3 and M-PEI6 for AR were 31.21 and 52.47 mg g⁻¹. The difference in the adsorption amounts of each membrane for the corresponding contaminant was mainly caused by the difference of the molecular weights of the contaminants; the adsorption capacities and selectivity of membranes also contribute to the difference in the adsorption amounts of each membrane.

Fig. 5 The adsorbed amounts per unit mass of GO/PES membranes for MB, RGO/PES membranes for BPA, and GO@PEI/PES membranes for AR at different time intervals (a); applications of the pseudo-first-order adsorption model (b), the pseudo-second-order adsorption model (c) and the intraparticle diffusion model (d).

In this study, the kinetics of the adsorption processes were analyzed using pseudo-first-order, pseudo-second-order and the intraparticle diffusion models. The pseudo-first-order kinetic model can be written in the following form:⁴⁴

ln(q_e − q_t) = ln q_e − k₁t (4)

where q_t is the adsorption amounts at time t (mg g⁻¹); q_e is the adsorption amounts at the equilibrium (mg g⁻¹); k₁ is the rate constant of pseudo-first-order equation (min⁻¹). The plot of ln(q_e − q_t) against t should give a straight line with slope −k₁ and intercept ln q_e, as seen in Fig. 5b. The parameters are shown in Table S1 (ESI†). The correlation coefficients (r²) of the GO/PES and GO@PEI/PES membranes were higher than 0.98, which meant that the adsorption process fitted the pseudo-first-order model. For the RGO/PES membranes, however, the results indicated that the adsorption process did not agree well with this model.

The pseudo-second-order equation can be represented in the following form:³⁰

where k₂ is the pseudo-second-order rate constant (g mg⁻¹ min⁻¹); q_t, q_e and t are the same as defined in the pseudo-first-order equation. From the slope and intercept of the plot of t/q_t versus t which are shown in Fig. 5c, the rate constant (k₂) and the equilibrium adsorption capacity (q_e) could be obtained. All the correlation coefficients (r²) were higher than 0.98 (seen in Table S1†), which meant that the contaminants adsorption on the prepared membranes agreed with the pseudo-second-order model. The calculated values of q_e were smaller than the experimental ones in pseudo-first-order equation (except M-PEI6) and larger than those in pseudo-second-order equation. By comparison, the values of q_e(cal) in pseudo-second-order model were more in accordance with the experimental values.

The successive process of contaminants diffusion through the boundary layer, intraparticle diffusion and adsorption on the surfaces of adsorbents could be described by the intraparticle diffusion model.²⁶ The equation could be written in the following form:

q_t = k_pt^1/2 + C (6)

where k_p is the intraparticle diffusion rate constant (mg g⁻¹ min^−1/2) and C represents the intercept related to the adsorption steps (mg g⁻¹); q_t has the same meaning as that in eqn (4). Fig. 5d shows the multilinear plots of intraparticle diffusion processes, indicating two steps have taken place: the first step represents the film diffusion stage and the second step represents the intraparticle diffusion stage. The slope of the linear part of each cure could give the rate constants, and the intercepts C could be obtained from the extrapolation of the first step in the curves to the time axis (seen in Table S2, ESI†). Most of the correlation coefficients (r²) were higher than 0.99, which meant that the adsorption process fitted this model. The rate constants of the first step (k₁) were larger than those of the second step (k₂), indicating film diffusion was a rapid process while the intraparticle diffusion was a slow process. The intercepts of the GO/PES membranes and GO@PEI/PES membranes were positive, which indicated that the boundary layer in the membranes promoted the intraparticle diffusion; instead, the intercepts of the RGO/PES membranes were negative, indicating an impeding effect of the membranes to the diffusion process.

3.4. Adsorption isotherms

The relationships of the equilibrium adsorption capacities with the initial concentrations of the contaminants are shown in Fig. 6a. With the increase of the initial concentrations of each contaminant, the adsorbed amounts of each membrane increased, and the increases of the adsorbed amounts began to flatten under high initial concentrations. When the adsorbed amounts were no longer increased with the increase of initial concentrations, the values of them were 32.57 and 43.02 mg g⁻¹ for MB removal by M-GO3 and M-GO6, respectively; 14.46 and 20.95 mg g⁻¹ for BPA removal by M-RGO3 and M-RGO6, respectively; and 36.41 and 62.90 mg g⁻¹ for AR removal by M-PEI3 and M-PEI6, respectively.

Fig. 6 The adsorbed amounts per unit mass of GO/PES membranes for MB, RGO/PES membranes for BPA and GO@PEI/PES membranes for AR with the different initial concentrations (a); application of Langmuir adsorption isotherm (b).

The adsorption isotherm can provide information about the adsorption behavior, the surface properties of adsorbent and the design of adsorption systems.⁴⁵ In this study, two adsorption isotherm models (Langmuir and Freundlich) were used to investigate the adsorption process.

The Langmuir adsorption isotherm assumed that the uptake of adsorbate molecules occurred on a homogenous surface with a finite number of adsorption sites by monolayer adsorption without any interaction between adsorbed molecules.⁴⁴ Once a site was occupied by adsorbate molecule, no further adsorption could occur at the site. The surface would reach the saturation point and the maximum adsorption of the surface would be achieved. The Langmuir equation was obtained as:⁴⁶

where C_e is the equilibrium concentration (mg L⁻¹); q_e is the adsorption amounts at the equilibrium (mg g⁻¹); q_max is the maximal adsorption capacities of the membranes (mg g⁻¹); k_L is the Langmuir adsorption constant. Fig. 6b shows the curves of contaminants adsorption on the membranes plotted with Langmuir model. The equilibrium concentration C_e and the maximal adsorption capacities q_max could be worked out by the slopes and the intercepts of the curves (seen in Table S3, ESI†). All the values of the correlation coefficients (r²) were higher than 0.99, which indicated that the adsorption data fit better according to the Langmuir isotherm well, and the adsorption processes were monolayer adsorption. The computational maximal adsorbed amounts of M-GO3 and M-GO6 for MB were 33.48 and 43.86 mg g⁻¹, respectively; those of M-RGO3 and M-RGO6 for BPA were 16.14 and 21.63 mg g⁻¹, respectively; and those of M-PEI3 and M-PEI6 for AR were 36.94 and 63.05 mg g⁻¹, respectively. These results were very close to the experimental values.

The Freundlich model is an empirical model based on multilayer adsorption on heterogeneous surface.⁴⁴ In this study, all the values of the correlation coefficients (r²) for the Freundlich model were less than 0.90, which indicated that the adsorption processes of the membranes did not fit this model (the fitting results were not given).

3.5. Effect of temperature and thermodynamic studies

The effect of temperature on the adsorbed amounts of the prepared membranes is shown in Fig. 7a. The results indicated that with the increase of temperature, the adsorbed amounts of all the membranes decreased, indicating an exothermic nature of the adsorption processes. The thermodynamic parameters, ΔG⁰ (standard free energy change), ΔH⁰ (enthalpy change) and ΔS⁰ (entropy change) were calculated to evaluate the feasibility and nature of the adsorption processes. The values of ΔH⁰ and ΔS⁰ were calculated from the slopes and intercepts of the plots of ln K_c versus 1/T (as shown in Fig. 7b) by using the following equation:⁴⁷

Fig. 7 The adsorbed amounts per unit mass of GO/PES membranes for MB, RGO/PES membranes for BPA and GO@PEI/PES membranes for AR in different adsorption temperatures (a); the fitting results of eqn (8) (b).

The ΔG⁰ was calculated from the following relation:

ΔG⁰ = ΔH⁰ − TΔS⁰ (9)

where R is the gas constant (8.314 kJ mol⁻¹); T is the Kelvin temperature (K); K_c is the standard thermodynamic equilibrium constant defined by q_e/C_e. The correlation coefficients (r²) of the fitting results and the values of ΔH⁰, ΔS⁰, and ΔG⁰ are given in Table S4 (ESI†). The values of r² were higher than 0.98, indicating that the adsorption data under various temperatures fitted the equation well. The values of ΔH⁰ were negative, indicating the exothermic processes of adsorption. The negative values of ΔS⁰ reflected the decreased randomness at the solid–liquid interface during the adsorption processes. The negative values of ΔG⁰ indicated that the adsorption was a spontaneous process, and the values of ΔG⁰ became more negative with the decrease in the temperature, indicating that lower temperature facilitated the adsorption processes.

3.6. Dynamic adsorption studies

In the practical operation of waste water filtration, the removal of contaminants by membranes is always a dynamic adsorption process. For the dynamic adsorption measurement, the prepared membrane was placed in a dead-end ultrafiltration cell, and the aqueous solution of adsorbate was passed the cell continuously. The system was said to reach the breakthrough point when the concentration of the filtered solutions rose to an appreciable value for the first time, and this time was the so-called breakthrough time (t_b). After t_b, the solute concentration in the effluent rose rapidly, and the exhausted point would appear when the concentration reached the initial value. The time of the emergence of exhausted point was called exhausted time (t_e).⁴⁸ In this study, the breakthrough point time and the exhausted point were considered at 0.1 and 0.9 (the ratio of the concentration at time t (C_t) to the initial concentration (C_i)), respectively. The adsorbed amounts of the membranes were calculated by the following equation:⁴⁸where q_e is the adsorbed amount of the membrane (mg g⁻¹); Q and W represent volumetric flow rate (mL min⁻¹) and the weight of the membranes (g); f(t) is the function representing the curve of C_t/C_i.

All the prepared membranes (except M-PES) were tested with their corresponding contaminants. The results are shown in Fig. 8a–c, and the parameters are shown in Table 2. One layer of membrane was applied in each test, the initial concentrations of each contaminant were 100 μmol L⁻¹, and the feed flow rates of the solutions were 0.3 g min⁻¹. Comparing with the adsorption column which was usually applied in the dynamic adsorption studies, the thickness of one layer of the membrane was extremely thin, so the solutions could pass through it quickly. As a result, the breakthrough points of each membrane were appeared after a short time. For each kind of membranes, the exhausted time increased with the increase of the GO-based materials contents, similarly, the adsorbed amounts per unit mass of membranes increased as well.

Fig. 8 Breakthrough curves for the adsorption of MB onto M-GO3 and M-GO6 (a), BPA onto M-RGO3 and M-RGO6 (b), and AR onto M-PEI3 and M-PEI6 (c), one layer of membrane is applied, the initial concentration is 100 μmol L⁻¹, the flow rate is 0.3 g min⁻¹. Breakthrough curves for the adsorption of MB onto M-GO6, BPA onto M-RGO6 and AR onto M-PEI6 (d), two layers of membranes are applied, the initial concentration is 50 μmol L⁻¹, the flow rate is 0.2 g min⁻¹.

Table 2 The dynamic adsorption parameters. * one layer of membrane is applied, the initial concentration is 100 μmol L⁻¹, the flow rate is 0.3 g min⁻¹; ** two layers of membranes are applied, the initial concentration is 50 μmol L⁻¹, the flow rate is 0.2 g min⁻¹; — t_b is less than 10 minutes

	tb (min)	te (min)	qe (mg g^−1)
*M-GO3	10	80	57.15
*M-GO6	20	140	80.99
*M-RGO3	—	40	14.07
*M-RGO6	—	60	18.92
*M-PEI3	10	80	79.18
*M-PEI6	10	100	104.41
**M-GO6	180	440	135.66
**M-RGO6	60	240	43.28
**M-PEI6	120	360	172.85

---

Some parameters such as the amount of the adsorbent, the initial concentration and the feed flow rate of contaminant solution greatly influenced the results of the dynamic adsorption process. In order to obtain the typical result (the breakthrough curves would have an S-shape), two layers of membranes were applied in the cell for the dynamic adsorption testing. The membranes with high GO-based materials contents (M-GO6, M-RGO6, and M-PEI6) were chosen because of their higher adsorption capacities and water fluxes. The initial concentration and the feed rate of each contaminant were adjusted to 50 μmol L⁻¹ and 0.2 g min⁻¹, respectively. Fig. 8d exhibits obvious S-shape breakthrough curves of each adsorption result. As seen from Table 2, the breakthrough times, the exhausted times and the adsorbed amounts significantly increased. The dynamic adsorption results indicated that the prepared membranes could be used for the field of wastewater filtration.

4. Conclusion

In this work, three kinds of GO-based polyethersulfone (PES) hybrid membranes with different adsorption characteristics were fabricated by liquid–liquid phase separation technique, i.e., GO/PES membranes, RGO/PES membranes and GO@PEI/PES membranes. The GO-based materials could be deposited in the PES membranes successfully. The adsorption tests on three chemicals, i.e., cationic dye methylene blue (MB), endocrine disruptor bisphenol A (BPA) and anionic dye amaranth (AR) were performed and the results indicated that the prepared membranes showed good adsorption capacities and selectivity for their corresponding contaminant. The adsorption kinetics studies indicated that the adsorption processes fitted the pseudo-first-order kinetic model and intraparticle diffusion model. The results of thermodynamics and isotherm studies indicated that the adsorption processes were exothermic and fitted with Langmuir isotherm well. In addition, the dynamic adsorption results indicated that the prepared membranes could be used for the field of wastewater filtration. The preparation of GO-based PES hybrid membranes herein could open up a route for the application of GO as adsorbent for broad water pollutants removal.

Acknowledgements

This work was financially sponsored by the National Natural Science Foundation of China (No. 51225303 and 51433007), the State Key Laboratory of Polymer Materials Engineering (Grant No. sklpme2015-1-03), and the Sichuan Province Youngth Science and Technology Innovation Team (No. 2015TD0001). We would also thank our laboratory members for their generous help, and gratefully acknowledge the help of Ms Hui Wang, of the Analytical and Testing Center at Sichuan University, for SEM observation.

References

1. A. Bhatnagar and M. Sillanpää, Chem. Eng. J., 2010, 157, 277–296.
2. A. M. Vandenbroucke, R. Morent, N. de Geyter and C. Leys, J. Hazard. Mater., 2011, 195, 30–54.
3. M. Rafatullah, O. Sulaiman, R. Hashim and A. Ahmad, J. Hazard. Mater., 2010, 177, 70–80.
4. A. Demirbas, J. Hazard. Mater., 2009, 167, 1–9.
5. V. Maroga Mboula, V. Hequet, Y. Andres, L. M. Pastrana-Martínez, J. M. Doña-Rodríguez, A. M. Silva and P. Falaras, Water Res., 2013, 47, 3997–4005.
6. B. Kasprzyk-Hordern, R. M. Dinsdale and A. J. Guwy, Water Res., 2009, 43, 363–380.
7. S. Rovani and M. T. Censi, J. Hazard. Mater., 2014, 271, 311–320.
8. F. Cao, P. Bai, H. Li, Y. Ma, X. Deng and C. Zhao, J. Hazard. Mater., 2009, 162, 791–798.
9. S. M. Alshehri, M. Naushad, T. Ahamad, Z. A. ALOthman and A. Aldalbahi, Chem. Eng. J., 2014, 254, 181–189.
10. B. O. Yoon, S. Koyanagi, T. Asano, M. Hara and A. Higuchi, J. Membr. Sci., 2003, 213, 137–144.
11. C. Zhao, Q. Wei, K. Yang, X. Liu, M. Nomizu and N. Nishi, Sep. Purif. Technol., 2004, 40, 297–302.
12. V. Vimonses, S. Lei, B. Jin, C. W. Chow and C. Saint, Chem. Eng. J., 2009, 148, 354–364.
13. K. G. Bhattacharyya and S. S. Gupta, Adv. Colloid Interface Sci., 2008, 140, 114–131.
14. M. Doğan, M. Alkan, A. Türkyilmaz and Y. Özdemir, J. Hazard. Mater., 2004, 109, 141–148.
15. V. Singh, A. Sharma, D. Tripathi and R. Sanghi, J. Hazard. Mater., 2009, 161, 955–966.
16. S. Kawano, T. Kida, K. Miyawaki, Y. Noguchi, E. Kato, T. Nakano and M. Akashi, Environ. Sci. Technol., 2014, 48, 8094–8100.
17. D. Chen, H. Feng and J. Li, Chem. Rev., 2012, 112, 6027–6053.
18. F. Kim, L. J. Cote and J. Huang, Adv. Mater., 2010, 22, 1954–1958.
19. D. R. Dreyer, S. Park, C. W. Bielawski and R. S. Ruoff, Chem. Soc. Rev., 2010, 39, 228–240.
20. T. Liu, Y. Li, Q. Du, J. Sun, Y. Jiao, G. Yang, Z. Wang, Y. Xia, W. Zhang and K. Wang, Colloids Surf., B, 2012, 90, 197–203.
21. Y. Sun, S. Yang, G. Zhao, Q. Wang and X. Wang, Chem.–Asian J., 2013, 8, 2755–2761.
22. G. Zhao, J. Li, X. Ren, C. Chen and X. Wang, Environ. Sci. Technol., 2011, 45, 10454–10462.
23. J. Zhao, Z. Wang, J. C. White and B. Xing, Environ. Sci. Technol., 2014, 48, 9995–10009.
24. V. Chandra, J. Park, Y. Chun, J. W. Lee, I.-C. Hwang and K. S. Kim, ACS Nano, 2010, 4, 3979–3986.
25. Y. Lin, S. Xu and J. Li, Chem. Eng. J., 2013, 225, 679–685.
26. F. Liu, S. Chung, G. Oh and T. S. Seo, ACS Appl. Mater. Interfaces, 2012, 4, 922–927.
27. H. Sun, Z. Xu and C. Gao, Adv. Mater., 2013, 25, 2554–2560.
28. Z. Sui, Y. Cui, J. Zhu and B. Han, ACS Appl. Mater. Interfaces, 2013, 5, 9172–9179.
29. C. S. Cheng, J. Deng, B. Lei, A. He, X. Zhang, L. Ma, S. Li and C. Zhao, J. Hazard. Mater., 2013, 263, 467–478.
30. X. Zhang, C. Cheng, J. Zhao, L. Ma, S. Sun and C. Zhao, Chem. Eng. J., 2013, 215, 72–81.
31. Q. Du, J. Sun, Y. Li, X. Yang, X. Wang, Z. Wang and L. Xia, Chem. Eng. J., 2014, 245, 99–106.
32. F. Jin, W. Lv, C. Zhang, Z. Li, R. Su, W. Qi, Q.-H. Yang and Z. He, RSC Adv., 2013, 3, 21394–21397.
33. J. Yin and B. Deng, J. Membr. Sci., 2015, 479, 256–275.
34. S. Sun, J. Hunag and C. Zhao, Sep. Purif. Rev., 2011, 40, 312–337.
35. D. Li, M. B. Mueller, S. Gilje, R. B. Kaner and G. G. Wallace, Nat. Nanotechnol., 2008, 3, 101–105.
36. C. Cheng, S. Li, S. Nie, W. Zhao, H. Yang, S. Sun and C. Zhao, Biomacromolecules, 2012, 13, 4236–4246.
37. L. Yang, B. Tang and P. Wu, J. Mater. Chem. A, 2014, 2, 18562–18573.
38. Z. Ding, C. Qian-Yi and H. Bao-Hang, Carbon, 2011, 49, 3920–3927.
39. C. Chong, M. Lang, R. Jian, L. Lulu, Z. Gaofeng, Y. Qingguo and Z. Changsheng, Chem. Eng. J., 2011, 171, 1132–1142.
40. C. Zhao, J. Xue, F. Ran and S. Sun, Prog. Mater. Sci., 2013, 58, 76–150.
41. W. Zhao, Q. Mou, X. Zhang, J. Shi, S. Sun and C. Zhao, Eur. Polym. J., 2013, 49, 738–751.
42. W. Zhang, C. Zhou, W. Zhou, A. Lei, Q. Zhang, Q. Wan and B. Zou, Bull. Environ. Contam. Toxicol., 2011, 87, 86–90.
43. Y. Chen, L. Chen, H. Bai and L. Li, J. Mater. Chem. A, 2014, 2, 13744.
44. J. Xu, L. Wang and Y. Zhu, Langmuir, 2012, 28, 8418–8425.
45. J. Fu, Z. Chen, M. Wang, S. Liu, J. Zhang, J. Zhang, R. Han and Q. Xu, Chem. Eng. J., 2015, 259, 53–61.
46. E. N. El Qada, S. J. Allen and G. M. Walker, Chem. Eng. J., 2008, 135, 174–184.
47. Y. Glocheux, M. Méndez Pasarín, A. B. Albadarin, C. Mangwandi, F. Chazarenc and G. M. Walker, Asia-Pac. J. Chem. Eng., 2014, 9, 929–940.
48. A. A. Attia, B. S. Girgis and N. A. Fathy, Dyes Pigm., 2008, 76, 282–289.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra20243d

---