Selective removal of anionic dyes using poly(N,N-dimethyl amino ethylmethacrylate) functionalized graphene oxide

Abstract

An in situ polymerization strategy was used to functionalize graphene oxide (GO) with poly(N,N-dimethyl amino ethylmethacrylate) (PDMAEMA) for the selective removal of anionic dyes. Various characterization methods demonstrate that PDMAEMA-grafted GO (GO-PDMAEMA) was successfully synthesized, and the high PDMAEMA content of 68.5% in GO-PDMAEMA changed the zeta potential significantly from −36.5 (GO) to 41.5 (GO-PDMAEMA). This change in the charge of GO-PDMAEMA greatly increased the adsorption capacities for anionic dye orange G (OG) compared to the pristine GO. The maximum adsorption capacity for anionic OG dye based on the Langmuir model is 609.8 mg g⁻¹. The adsorption mechanism is believed to be a consecutive process of intra-particle diffusion and surface adsorption, with electrostatic interactions as the key driving force. The GO-PDMAEMA nanocomposite also showed excellent regeneration capacity and selectivity towards the separation of various anionic dyes (i.e. OG, Eosin yellow and Congo red) from an aqueous dye mixture. In conclusion, our method offers a promising strategy for developing new anionic dye adsorbents.

---

1. Introduction

Due to industrialization and urbanization, water pollution has become a severe social problem worldwide. Pollutants found in water bodies include organic dyes, heavy metal ions, toxic elements, and pesticides among other organics and inorganics.¹ To protect precious water resources from being contaminated, sewage containing pollutants must be treated before final disposal. Of all the methods proposed, adsorption technology is globally regarded as the most promising method for wastewater treatment due to its versatility, wide applicability and economic feasibility.

With good dispersion,^2,3 high specific surface,⁴ and abundant functional groups,⁵ graphene oxide (GO) is commonly regarded as a kind of ideal absorbents for pollutant removal. Recent literature indicated that GO can be used as a promising absorbent for toxic ions such as Cu²⁺ (ref. 2) and Cr³⁺ (ref. 6) and cationic dyes such as methyl green⁷ and methyl blue (MB).⁸ However, almost all previous reports can only deal with the positive ions or cationic dyes, because the negatively charged GO surface produces strong electrostatic repulsion between GO and anionic dyes or negative ions.⁹ For instance, Ramesha et al.⁹ compared the removal efficiency of GO for both cationic and anionic dyes and found that GO can remove cationic methylene blue and methyl violet effectively. However, due to the strong electrostatic repulsion between GO and anionic orange G (OG), GO shows much weaker removal ability for OG than that of cationic methylene blue or methyl violet.⁹ Thus, to expand GO application especially for the removal of anionic dyes or negative ions, further functionalization of GO by introducing cationic groups is necessary.

The GO functionalization has recently been explored by free radical polymerization, atom transfer radical polymerisation (ATRP) and in situ polymerization.^10–13 For instance, free radical polymerisation was explored to graft polyacrylic acid (PAA) and polyacryl amide (PAM) onto graphene sheets in solution phase.^10,11 However, the free radicals produced in the solution can easily produce homo-polymers and result in low graft efficiency. Gil et al.,¹⁴ on the other hand, utilised ATRP to graft poly(methyl methacrylate) (PMMA) chains onto GO surface. It is commonly known that the ATRP technique is a highly efficient method to eliminate the chain transfer and thus capable of controlling the polymerization process. However, ATRP is a complex preparation process, requiring multi-step reaction for ATRP initiator synthesis.¹⁴ More recently, a series of homo-polymers were grafted onto GO via Ce(IV) salts initiated redox polymerisation at moderate temperatures,^13,15 where the free radicals were formed via the hydroxyl group oxidation on GO surfaces in presence of Ce(IV) salts. Compared to other approaches, this method can be easily applied without any side-reaction interfered.

In the current study, a new GO derivative graphene oxide-poly[2-(N,N-dimethyl amino) ethyl methacrylate] (GO-PDMAEMA) with positive charge surface will be synthesized using Ce(IV) salt initiated polymerisation strategy.¹⁵ The tertiary amine groups within the PDMAEMA will then be protonated under acid environment. Those protonated amines can transform the negative surface of GO into positive surface. It is anticipated that this new material can be used for the removal of anionic dyes or negative ions through electrostatic attraction. To demonstrate this, the adsorption behaviors of the as-synthesized GO-PDMAEMA will be systematically evaluated using anionic OG and cationic MB as models. Its regeneration and selectivity ability will also be systematically evaluated.

2. Experimental

2.1. Materials

Graphite (particle size < 45 μm), OG, MB, cerium(IV) ammonium nitrate (CAN), Eosin yellow (EY) and N,N-dimethylaminoethyl methacrylate (DMAEMA) were all analytical reagent and purchased from Sigma Aldrich.

2.2. Synthesis of GO-PDMAEMA

The GO was synthesised used the same method as reported,¹⁵ which was then dispersed in deionised water and purified by dialysis in deionised water for 6 days. The outside deionised water was changed twice every day. After that, the GO dispersion was transferred to a beak and exfoliated by using a probe sonicator (Shengxi Instrument Company, Shanghai, China) with a 60% of amplitude (240w). The as-obtained GO was concentrated to obtain 0.4 wt% dispersion after exfoliation and dialysis. To synthesise GO-PDMAEMA, 0.02 mol DMAEMA was firstly dissolved in 50 mL 0.4 mol L⁻¹ HCl solution, and 500 mL 0.4% GO was then added. The mixed solution was put in a 25 °C water bath and degassed for 20 min using slow nitrogen flow. After that, 10 mL CAN solution (1.1 g CAN in 10 mL deionized water) was added dropwise. The reaction was then maintained at 50 °C for 120 min under stirring. It was found that the viscosity of the mixture was increased during the polymerization. Subsequently, the suspension was centrifuged and rinsed with deionized water for five times to remove unreacted monomers, residual ions and trace homo-polymers.

2.3. Characterizations

GO-PDMAEMA and GO morphologies were captured via a Carl Zeiss AG scanning electron microscopy (SEM) Supra 55VP at a voltage of 5 kV. Raman micro spectroscopic measurements were conducted via a RenishawInVia Raman Microspectrometer (Renishawplc, Gloucestershire, UK), equipped with a 633 nm laser.¹⁵ ATR-FTIR spectroscopic technique was recorded using a Bruker Vertex-70 FTIR spectrometer with 4 cm⁻¹ spectral resolution. ¹³C and ¹H solid NMR experiments were conducted on a BrukerAvanceIII 300 MHz wide bore NMR spectrometer (¹H Lamor frequency of 300.13 MHz) with a 4 mm double resonance Magic Angle Spinning (MAS) probe head using the same procedure before.¹⁵

2.4. Adsorption kinetics and isothermal adsorption

Dye concentration was determined using UV-vis spectrophotometer (Shimadzu UV-260) based on the maximum absorbance wavelengths (λ_max). The λ_max for OG and MB was used as 478 (ref. 16) and 662 nm (ref. 17) at 25 °C using deionised water as solvent, respectively. By correlating absorbance A to concentration (mg L⁻¹), calibration plots were established based on Beer–Lambert's law. In all the following adsorption experiments, deionised water was chosen to dissolve various dyes. For adsorption kinetics study, 0.0200 ± 0.0001 g of GO-PDMAEMA (or GO) was added into a vial containing 20.00 ± 0.01 g dye solution at 25 °C for a predetermined time interval. For the isothermal adsorption study, 0.0200 ± 0.0001 g of GO-PDMAEMA (or GO) adsorbent was slowly transferred into a 100 mL vial containing 20.00 g model dye solutions and kept for consecutive 3 days at 25 °C to reach equilibrium. The calibration curves obtained was used to analyse the absorbance change of dye solution which was recorded spectrally.

The adsorbent capacity (mg g⁻¹) (q_t) of GO-PDMAEMA (or GO) was analysed with the mass balance eqn (1):

where v is the volume of dye solution, C₀ and C_e are the initial and equilibrium concentrations of dye, respectively.

2.5. Regeneration & selective adsorption

Desorption experiments were conducted in a vial to recover the adsorbed anionic dye OG and to regenerate the used GO-PDMAEMA for further OG removal. For desorption experiments, 0.1000 ± 0.0001 g GO-PDMAEMA was added into 100.00 ± 0.01 g dye solution (200 mg L⁻¹) at 25 °C for three days. After that, GO-PDMAEMA loaded with dyes was separated and put into 10.0 mL NaOH solution (0.001 mol L⁻¹). After desorption, the GO-PDMAEMA was rinsed using deionized water and regenerated in 10 mL 0.001 mol L⁻¹ HCl solution for 10 minutes. It was then separated and washed with deionized water. It was ready to be reused for the adsorption experiment in the next cycles.

Selective removal of anionic dyes was performed from a mixture of cationic MB and anionic dyes (i.e. OG, EY and CR) under same concentration. The GO-PDMAEMA dosage was fixed at 20 mg/20 mL in the mixed solution at a temperature of 313 K and the absorbance variation was monitored by UV-vis spectrometer with respect to contact time.

3. Results and discussion

3.1. Characterization

The morphology of as-prepared GO and GO-PDMAEMA was firstly compared via SEM investigation. As shown in Fig. 1(a), the as-synthesized GO is of typical layered structure. After the functionization, the GO-PDMAEMA becomes much thicker and rougher with some particles formed on the GO sheet surface (Fig. 1(b)). Those PDMAEMA particles also incline to aggregate together, similar to previous investigation.¹²

Fig. 1 SEM images. (a) GO and (b) GO-PDMAEMA.

To investigate the chemical structure of GO-PAMAEMA, ATR-FTIR, Raman, NMR, elemental analysis techniques were applied respectively. For the origin GO, characteristics stretching vibration of C=C (1628 cm⁻¹), C=O (1728 cm⁻¹) and O–H (3375 cm⁻¹) groups were clearly indicated in Fig. 2(a).¹⁵ After the functionization, new peaks centered at 1462, 1381 and 1149 cm⁻¹ are emerged in the ATR-FTIR spectrum of the GO-PDMAEMA (Fig. 2(a)), which are attributed to the –CH₂ bending, CH₃ symmetric scissors and C–N stretching respectively.¹⁸ In addition, the C=O band in GO-PDMAEMA, which is much stronger than that of GO, was shifted to 1722 cm⁻¹. All these changes clearly show that PDMAEMA has been successfully introduced to the surface of GO. The stretching vibration of O–H was shifted from 3375 cm⁻¹ (GO) to 3362 cm⁻¹ (GO-PDMAEMA), indicating that there are interactions between GO and PDMAEMA chains. The two broad peaks at around 1600 and 1350 cm⁻¹ are assigned to G (graphite) and D (defect) bands, respectively (Fig. 2(b)). G band represents the characteristic signals of crystalline graphite, while the D band is due to the existence of amorphous carbon (i.e. diamond sp³ network).¹⁹ It seems that there are no clearly observed differences between GO and GO-PDMAEMA. To identify the changes of D and G bands quantitatively, the integrated area of D band (A_D) and G band (A_G) were further calculated via the deconvolution method. The results in Fig. 2(c) shown that the single broad G band observed in Fig. 2(b) is in fact three combined components G, G′ and G′′. In contrast, the D band is composed by two components D and D′. The A_D/A_G values obtained for GO and GO-PDMAEMA samples were found to be 1.56 and 1.52, respectively. The slightly lower A_D/A_G value for the GO-PDMAEMA shows declined defect domains and thus increased graphitic domains in the as-synthesized GO-PDMAEMA composite compared to the pristine GO, which may be due to the oxidation of GO functional groups in GO-PDMAEMA composite under the existence of oxidative cerium(IV) ions or the additional interactions of PDMAEMA chains.¹⁵

Fig. 2 (a) Comparison of (a) FTIR spectra feature, (b) normalized Raman spectra feature and (c) Raman spectral deconvolution of GO and GO-PDMAEMA.

Although the Raman and ATR-FTIR analyses indicated that PDMAEMA chains were incorporated within the GO-PDMAEMA composites, the structure and dynamics of the grafted polymer chain at molecular level was required to be further investigated. Therefore, solid-state NMR spectroscopy was recorded in Fig. 3. As a kind of robust techniques, solid-state NMR spectroscopy can observe both the molecular structure and the dynamics of solid polymers.¹⁵ In polymer solids, the ¹H NMR spectral line width is generally dominated by proton homonuclear dipole–dipole interactions. Molecular motions in the frequency range of kHz to a few hundreds of kHz would average out the dipole–dipole interactions and thereby lead to line narrowing. In most cases, the polymer chain motion falls into this frequency range, and thus the ¹H NMR spectral line width is closely related to the mobility of polymer chain and provides very useful information for molecular dynamics.²⁰

Fig. 3 NMR spectra of GO and GO-PDMAEAM: (a) static ¹H NMR and (b) ¹³C NMR.

It is interesting to notice that there is a negligible ¹H signal in the raw GO. However a strong ¹H NMR signal is presented in spectrum of the GO-PDMAEMA (Fig. 3(a)), indicating that the strong proton signals is attributed to the grafted PDMAEMA chains. The GO-PDMAEMA sample has a line width of about 46 kHz. The proton line width of 46 kHz suggests a relatively slow polymer chain dynamics, which is most probably caused by the covalent bonding between the polymer chain and GO surface. Physical adsorption may be another possibility to impose restriction on the polymer chain, however this is not likely in this study considering the samples had been centrifuged and rinsed with deionized water for five times. Therefore, the ¹H NMR results suggest that chemical restrictions for PDMAEMA chains are existed in GO-PDMAEMA composite.¹⁵

The ¹³C NMR spectra of the GO and the GO-PDMAEMA are shown in Fig. 3(b). Two broad peaks are observed at 129 and 66 ppm in the pristine GO sample, as a representation of sp² and sp³ hybridized ¹³C,¹⁵ respectively. The very broad peak centered at 129 ppm shows that there are many aromatic carbons with a broad distribution of chemical shift in the GO, indicating the existence of the large π-conjugate system.²¹ Furthermore, it was suggested that the peak at 66 ppm is a band overlapped from two peaks normally centred at 59.7 and 69.6 ppm, respectively, assigned to the epoxide ¹³C and ¹³C–OH.²² Apart from these two peaks (i.e. 129 and 66 ppm), there are seven different carbon sites in PDMAEMA molecules, but only five new peaks are observed in the GO-PDMAMEA and are all assigned to the characteristic carbon atoms in PDMAEMA (Fig. 3(b)).²³

To further quantify the PDMAEMA composition, elemental analysis results were presented in Table 1. As the N content is 6.1 mol% in GO-PDMAEMA, the polymer content in the GO-PDMAEMA composite is about 68.5 wt%. Zeta potential is an important parameter for absorbents. According to Table 1, the zeta potential for pristine GO is −36.5, indicating that GO can be an effective cationic dye absorbent. After the functionization, the zeta potential for GO-PDMAEMA is changed from negative to positive (41.5), indicating that GO-PDMAEMA can be used as an effective anionic dye absorbent.

Table 1 Elemental analysis and zeta potential results

Samples	Element analysis (%)			Zeta potential (mV)
	C	H	N
GO	45.9%	2.8%	0.0%	−36.5 ± 1.1
GO-PDMAEMA	47.1%	4.7%	6.1%	41.5 ± 1.3

---

3.2. Adsorption kinetics

The dye uptake is determined by the contact time between the absorbent and the dye solution and the diffusion processes.²⁴ Fig. 4(a and b) shows the dependence of OG and MB adsorbed on contact time. The OG or MB adsorption increased sharply initially, and then reached an equilibrium value at around 1000 min. For OG removal, adsorption capacity of GO-PDMAEAM is much higher than that of GO under whole adsorption process (Fig. 4(a)). On the contrary, the MB adsorption capacity of GO is much higher than that of GO-PDMAEAM during the whole adsorption process (Fig. 4(a)). This indicates that the anionic dye removal ability of the as-synthesized GO-PDMAEAM is significantly improved.

Fig. 4 Effect of contact time on the adsorption of OG (a) and MB (b); linear dependence of ln(q_e − q_t) on t for the adsorption of OG (c) and MB (d) based on the pseudo-first-order model; linear dependence of ln(q_e − q_t) on t for the adsorption of OG (e) and MB (f) based on the pseudo-second-order model; linear dependence of q_t on t for the adsorption of OG (g) and MB (h) based on the pseudo-second-order model.

The adsorption kinetics of OG or MB can be evaluated by analyzing the experimental kinetic data with the pseudo-first-order²⁵ and pseudo-second-order kinetic^26,27 models. The linear pseudo-first-order^25,26 model follows,

ln(q_e − q_t) = ln q_e − k₁t (2)

where q_e (mg g⁻¹) is the adsorbed model dyes at equilibrium and q_t (mg g⁻¹) is the model dyes adsorbed at time t. k₁ (min⁻¹) is the adsorption rate constant for the pseudo-first-order model. And pseudo-second-order kinetic rate equation can be expressed as,²⁸where k₂ (mg g⁻¹ min⁻¹) is the adsorption rate constant for the pseudo-second-order kinetic model. For eqn (2), the equilibrium adsorption parameter q_e must be defined in advance. On the contrary, there is no need to know any parameters beforehand for eqn (3), as q_e, can be directly calculated. Thus, pseudo-second-order kinetic model is more reliable when the whole adsorption behaviors are predicted.

The linear log(q_e − q_t) vs. t and t/q_t vs. t were plotted in Fig. 4(c–f). The kinetic parameters obtained from liner fitting of these two models were given in Table S1.† The experimental data showed good compliance with both models as a high correlation coefficient value (R² ≥ 0.95) has been achieved. The calculated q_e,cal with the pseudo-second-order model is a little higher than that of experimental q_e,exp. However, the calculated q_e,cal from the pseudo-first-order model is much lower than the q_e,exp. The q_e,exp may be slightly lower than the real q_e due to equilibrium not fully reached. Therefore, it can be postulated that the pseudo-second-order kinetics model is more applicable than the pseudo-first-order model for the adsorption of the two models dyes.

Considering that pseudo-second-order kinetic model cannot determine the diffusion mechanism, an intra-particle diffusion model²⁹ was introduced to analyze the adsorption mechanism:

q_t = k_idt^0.5 + C (4)

where k_id (mg g⁻¹ min^−0.5) is a constant that describes intra-particle rate and C is the thickness between boundary layers.²⁴

The overall dye adsorption process can be divided into the following three stages. The first stage is called outer diffusion or boundary layer diffusion. In this stage dye molecules migrate from the bulk solution to the external surface of adsorbent. The inner diffusion is the second stage when the dye molecules are transported from the external surface of adsorbent to the pores or capillaries of the internal structure. It is also called intra-particle diffusion. In the last stage, dye molecules are adsorbed onto the active sites of internal and external surfaces of the adsorbent.³⁰ This step is fast and thus cannot be rate-controlled. Therefore the adsorption rate was controlled by either inner diffusion, outer diffusion, or inner and outer diffusions.³¹

The resulting kinetic plots based on intra-particle diffusion model was presented in Fig. 4(g and h) and the corresponding kinetic parameters were given in Table S1† as well. The high correlation coefficient value indicates that the experiment data are well compliant with the intra-particle diffusion model. Furthermore, the plot did not pass through the origin (C ≠ 0), showing that the thickness of the boundary layer, suggesting that surface adsorption may also play an important role during the dye uptake process in combination with intra particle diffusion model. Therefore surface adsorption and intra-particle diffusion takes place simultaneously during the whole adsorption process.³²

3.3. Adsorption isothermals

An adsorption isotherm describes the relationship between the adsorption capacities and the equilibrium sorbate concentration at constant temperature. Certain constants of adsorption isotherm models can well reveal the surface properties and affinity of the adsorbent. To optimize the whole adsorption system, it is of vital importance to analyse isotherm data by fitting them to various models. The Langmuir and Freundlich equations are the most widely used isotherm models.

In the Langmuir isotherm model, an activate point on the surface of the adsorbent only adsorbs one molecule. Therefore, the adsorbed layer is not thicker than one molecule, indicating that the sorption is homogeneous. It is assumed that no chemical reaction is involved, the Langmuir equation is expressed as³³

where C_e is liquid-phase concentrations of dye (mg g⁻¹) at the equilibrium, q_m represents maximal sorption capacity which corresponds to complete monolayer coverage, and K_l is Langmuir constant (L mg⁻¹) and is associated to the adsorption energy. The fundamental characteristic of Langmuir model is the separation factor R_L, and is given by:where C₀ is the initial dye concentration (mg dm⁻³).³⁴ R_L is dimensionless. If the sorption process is irreversible, R_L = 0; if favorable, 0 < R_L < 1; if unfavorable (R_L > 1); and when the process is linear, R_L = 1.

Another model, Freundlich isotherm, is generally used for non-ideal multilayer adsorption on unhomogeneous surfaces. And its linear form is:

where k_f is a capacity factor associated to the adsorption capacity ((mg g⁻¹)(L mg⁻¹)^1/n) of the materials. The n is adsorption parameter and represents the adsorption intensity. A high n indicates a strong interaction between adsorbent and adsorbate.³⁵

The experimental sorption isotherms were presented in Fig. 5(a and b). As expected, positive charged GO-PDMAEMA shows greatly increased adsorption capacities for anionic OG compared to the pristine GO at various OG concentration. The results analyzed from both Langmuir and Freundlich models are illustrated in Fig. 5(c–f), respectively, while the parameters computed with Langmuir and Freundlich models (eqn (5)–(7)) are presented in Table S2.† Similar to previous reports,^36,37 there are no clear differences between the correlation coefficient R² values of Langmuir and Freundlich models (all higher than 0.95). Thus it is assumed that both models can describe the whole adsorption process, indicating that multiple interactions are present in those systems with a tendency to saturate both the interaction sites and heterogeneous adsorption sites.^36,37 The maximum adsorption capacities of OG calculated by Langmuir model are 161.6 and 609.8 mg g⁻¹ for GO and GO-PDMAEMA, respectively, indicating the anionic dye adsorption capacities of GO-PDMAEMA is greatly improved. And the adsorption capacity of GO-PDMAEMA based on Langmuir model is significantly higher than that of industrial adsorbents such as activated carbon (64.93 mg g⁻¹)³⁸ and bagasse fly ash (18.80 mg g⁻¹).³⁸ The maximum adsorption capacity q_m of MB calculated by Langmuir model are 232.6 and 41.1 mg g⁻¹ for GO and GO-PDMAEMA, respectively. Moreover, the parameter k_F associated with adsorption capacity in Freundlich model shows the same order (i.e. k_{F(OG on GO-PDMAEMA)} > k_{F(MB on GO)} > k_{F(OG on GO-PDMAEMA)} > k_{F(MB on GO)}) as q_m in Langmuir model. Both these results show that GO-PDMAEMA possess improved removal ability of OG but decreased removal efficiency of MB. On the other hand, the values of n in Table S2† show that n_{(MB on GO)} > n_{(MB on GO-PDMAEMA)} > n_{(OG on GO-PDMAEMA)} > n_{(OG on GO)}, indicating that the bonding between MB and GO (or GO-PDMAEMA) is stronger than that between OG and GO-PDMAEMA. This may be due to the strong interaction between π electrons of benzene rings on GO (or GO-PDMAEMA) and π electrons in MB.^17,39 The relative weak bonding between OG and GO-PDMAEMA indicates that desorption and regeneration is feasible. Combining the high adsorption capacity (609.8 mg g⁻¹) and the weak bonding, GO-PDMAEMA is a very competitive and viable adsorbent for OG in practical application.

Fig. 5 Experimental sorption isotherms of OG (a) and MB (b); Langmuir isotherm models for sorption of OG (c) and MB (d); Freundlich isotherm models for sorption of OG (e) and MB (f); separation factor R_L for the adsorption of OG and MB (g).

Variation of separation factor R_L for the adsorption with the concentration of OG and MB is shown in Fig. 5(g), indicating that both MB and OG adsorption is more favorable at higher initial dye concentration.⁴⁰ Due to the strong interactions (or driving forces) between MB and GO (or GO-PDMAEMA), MB adsorption is spontaneous (R_{L(MB on GO)} = 0.01, R_{L(MB on GO-PDMAEMA)} = 0.22) even when MB concentration is as low as 20 mg L⁻¹. However, uptake of 20 mg L⁻¹ OG using GO is extremely hard as the OG adsorption on GO is nearly linear (R_{L(OG on GO)} is 0.98 when OG concentration is 20 mg L⁻¹, very close to 1). After the grafting, the R_{L(OG on GO-PDMAEMA)} value at 20 mg L⁻¹ OG solution is decreased greatly to 0.77, showing a much more favorable interaction. In conclusion, GO-PDMAEMA shows greatly improved adsorption capacity and appropriate separation ability for anionic dye OG.

3.4. Adsorption mechanism

Based on the zeta potential result, the positive GO-PDMAEAM composites can effectively bond the negative OG. To further confirm the interaction mechanism, the effect of salt concentration on OG adsorption capacity was investigated (Fig. 6) to examine the contribution of ionic strength. It reveals that the adsorption capacity reduced significantly when the outer ions were introduced. With the increase in ionic concentration (or ionic strength) of both salts, the corresponding adsorption capacity declined gradually. This phenomenon is due to the competitive interactions between anions (Cl⁻ and SO₄²⁻) and hydrated anionic dyes with the positively charged GO-PEMAEMA composite. Similar findings were also reported where cationically modified guar gum was used to remove anionic CR.⁴¹ The outer “ionic atmosphere” may also produce shielding effect on the charge of anionic dyes, leading to the decline of the adsorption rate.⁴² Nevertheless, as the ionic strength of SO₄²⁻ is much greater than that of Cl⁻, the adsorption capacity of GO-PEMAEMA in presence of SO₄²⁻ ions is much lower than that of Cl⁻ under the same concentration. This result suggests that the predominating interaction between GO-PDMAEMA nanocomposite and OG dye is electrostatic attraction.

Fig. 6 Effects of salt concentration on adsorption for OG at 25 °C.

To investigate the interactions between OG dyes and GO-PDMAEMA nanocomposite, the FTIR spectra of the OG dye, GO-PDMAEMA nanocomposite and nanocomposite with adsorbed OG were compared in Fig. 7, where a couple of characteristic band shifts were found for OG dyes loaded GO-PDMAEMA compared to the pure GO-PDMAEMA, e.g. –OH stretching from 3362 cm⁻¹ to 3381 cm⁻¹, C=C bond from 1630 cm⁻¹ to 1624 cm⁻¹, –CH₂ bending from 1462 cm⁻¹ to 1467 cm⁻¹ and C–N stretching from 1149 cm⁻¹ to 1142 cm⁻¹. However, there are no any observable changes for C=O bonds (1722 cm⁻¹). Those FTIR shifts further confirm that there are interactions between the active sites of cationic GO-PDMAEMA nanocomposite and anionic OG dyes.²⁴

Fig. 7 FTIR spectra of (a) OG, (b) GO-PDMAMEA and (c) OG absorbed to GO-PDMAMEA.

3.5. Recyclability and selectivity

Desorption experiments were conducted to investigate the suitability of reusing adsorbent in the solutions of predetermined respective pH 7 and 11. The observed desorption rate is 94.5% and 62.2% at pH 11 and 7 respectively, suggesting that ion exchange predominates in OG desorption process.²⁴ The possible desorption mechanism is presented in Scheme S1 (ESI†). To improve the regeneration efficiency, the GO-PDMAEMA used was then immersed in 0.01 mol L⁻¹ HCl solution for 20 minutes. It was found that the adsorption capacity of GO-PDMAEMA is as high as 76.4% of the original one after the 3rd cycle (Fig. 8(a)). The decrease in capacity may be due to the un-desorbed OG as well as the weight loss of GO-PDMAEMA during desorption and regeneration process. However, it can be concluded that GO-PDMAEMA showed good recyclability for the treatment of OG dye from aqueous solution.

Fig. 8 (a) Adsorption capacity percentage with desorption cycle; UV-vis spectra shows the selective removal of (b) mixture of MB and OG and (c) mixture of EY and MB.

Selective removal of anionic or cationic dyes from their mixture is very important for practical applications.^41,43 Fig. 8(b and c) present the UV-vis spectra of the mixture of anionic dyes (OG and EY) and cationic MB dye before and after adsorption at different time interval. The experimental results show that the characteristic peaks of the OG (at 478 nm) and EY (at 518 nm) were decreased greatly after adsorption while the change for the MB (662 nm) is very limited. Those experimental observations indicate that electrostatic interactions between anion OG (or EY) and positively charged GO-PDMAEMA is primarily responsible for the selective adsorption of anionic dyes from their mixture. The similar selectivity was also found in the mixture of MB and CR (Fig. S1(a)†), where (MB₁CR₁) complex (Fig. S1(b) and S2†) is formed in the aqueous environment.⁴⁴ This as-observed excellent selectivity further supports the feasibility of using GO-PDMAEMA as a competent adsorbent for selective removal of anionic dyes.

4. Conclusions

The GO-PDMAEMA nanocomposite was synthesized via in situ polymerization of DMAEMA on graphene oxide. FTIR, Raman spectroscopy, SEM, NMR and EA analyses proved the successful grafting of PDMAEMA. Due to the positive charge of GO-PDMAEMA, the as-synthesized GO-PDMAEMA demonstrates much higher adsorption capacity for anionic OG dye compared to the pure GO. The experimental data fit well with the pseudo-second-order model, and the adsorption isotherm followed both the Langmuir and Freundlich models. The maximum adsorption capacity of GO-PDMAEMA for anionic OG dye based on Langmuir model is 609.8 mg g⁻¹, which is much higher than that of the pristine GO (161.6 mg g⁻¹). The adsorption mechanism can be regarded as consecutive process of intra-particle diffusion and surface adsorption. Furthermore, it is found that the electrostatic attraction is responsible for the interactions between the active sites of anionic OG and cationic GO-PDMAEMA. Moreover, recycling studies confirmed that the GO-PDMAEMA nanocomposite could be used repeatedly for the adsorption–desorption of OG. Particularly, the GO-PDMAEMA nanocomposite shows excellent selectivity towards the separation of anionic dyes form the binary dye systems. Therefore, the as-fabricated GO-PDMAEMA nanocomposite has the potential to be used as a promising adsorbent for selective removal of anionic dyes from industrial wastewater.

References

1. S. Chowdhury and R. Balasubramanian, Adv. Colloid Interface Sci., 2014, 204, 35–56.
2. S. T. Yang, Y. L. Chang, H. F. Wang, G. B. Liu, S. Chen, Y. W. Wang, Y. F. Liu and A. N. Cao, J. Colloid Interface Sci., 2010, 351, 122–127.
3. F. Kim, L. J. Cote and J. X. Huang, Adv. Mater., 2010, 22, 1954–1958.
4. Y. Pan, T. Wu, H. Bao and L. Li, Carbohydr. Polym., 2011, 83, 1908–1915.
5. S. Stankovich, D. A. Dikin, R. D. Piner, K. A. Kohlhaas, A. Kleinhammes, Y. Jia, Y. Wu, S. T. Nguyen and R. S. Ruoff, Carbon, 2007, 45, 1558–1565.
6. Y. Yuan, G. H. Zhang, Y. Li, G. L. Zhang, F. B. Zhang and X. B. Fan, Polym. Chem., 2013, 4, 2164–2167.
7. P. Sharma and M. R. Das, J. Chem. Eng. Data, 2012, 58, 151–158.
8. C. Wang, J. Zhou and L. Chu, RSC Adv., 2015, 5, 52466–52472.
9. G. K. Ramesha, A. Vijaya Kumara, H. B. Muralidhara and S. Sampath, J. Colloid Interface Sci., 2011, 361, 270–277.
10. J. F. Shen, Y. Z. Hu, C. Li, C. Qin, M. Shi and M. X. Ye, Langmuir, 2009, 25, 6122–6128.
11. G. Wang, X. Shen, B. Wang, J. Yao and J. Park, Carbon, 2009, 47, 1359–1364.
12. Y. F. Yang, J. Wang, J. Zhang, J. C. Liu, X. L. Yang and H. Y. Zhao, Langmuir, 2009, 25, 11808–11814.
13. L. Ma, X. Yang, L. Gao, M. Lu, C. Guo, Y. Li, Y. Tu and X. Zhu, Carbon, 2013, 53, 269–276.
14. G. Goncalves, P. A. A. P. Marques, A. Barros-Timmons, I. Bdkin, M. K. Singh, N. Emami and J. Gracio, J. Mater. Chem., 2010, 20, 9927–9934.
15. C. Li, H. Zhu, T. Hou, J. Vongsvivut, J. X. Dai, F. She and L. Kong, Chem. Eng. J., 2015, 263, 27–37.
16. M. Q. Cai, X. Q. Wei, Z. J. Song and M. C. Jin, Ultrason. Sonochem., 2015, 22, 167–173.
17. C. Li, M. She, X. She, J. Dai and L. Kong, J. Appl. Polym. Sci., 2014, 131, 9872–9881.
18. D. Roy, J. T. Guthrie and S. Perrier, Soft Matter, 2008, 4, 145–155.
19. M. A. Pimenta, G. Dresselhaus, M. S. Dresselhaus, L. G. Cancado, A. Jorio and R. Saito, Phys. Chem. Chem. Phys., 2007, 9, 1276–1290.
20. H. Zhu, R. Graf, G. Hou, Y. Zhao, D. Wang and H. W. Spiess, Macromol. Chem. Phys., 2010, 211, 1157–1166.
21. B. Li, Z. Guan, X. Yang, W. D. Wang, W. Wang, I. Hussain, K. Song, B. Tan and T. Li, J. Mater. Chem. A, 2014, 2, 11930–11939.
22. W. W. Cai, R. D. Piner, F. J. Stadermann, S. Park, M. A. Shaibat, Y. Ishii, D. X. Yang, A. Velamakanni, S. J. An, M. Stoller, J. H. An, D. M. Chen and R. S. Ruoff, Science, 2008, 321, 1815–1817.
23. S.-C. Chen, S.-W. Kuo, C.-S. Liao and F.-C. Chang, Macromolecules, 2008, 41, 8865–8876.
24. S. Ghorai, A. Sarkar, M. Raoufi, A. B. Panda, H. Schonherr and S. Pal, ACS Appl. Mater. Interfaces, 2014, 6, 4766–4777.
25. M. Dogan, M. Alkan, A. Türkyilmaz and Y. Özdemir, J. Hazard. Mater., 2004, 109, 141–148.
26. Y. S. Ho and G. McKay, Process Biochem., 1999, 34, 451–465.
27. Y. Zheng, Y. Liu and A. Wang, Chem. Eng. J., 2011, 171, 1201–1208.
28. J. Ma, Y. Zhuang and F. Yu, New J. Chem., 2015, 39, 9299–9305.
29. S. Song, Y. Ma, H. Shen, M. Zhang and Z. Zhang, RSC Adv., 2015, 5, 27922–27932.
30. F. Yu, Y. Q. Wu, X. M. Li and J. Ma, J. Agric. Food Chem., 2012, 60, 12245–12253.
31. F. Yu, S. N. Sun, S. Han, J. Zheng and J. Ma, Chem. Eng. J., 2016, 285, 588–595.
32. C. Ou, S. Zhang, J. Liu, J. Shen, Y. Liu, X. Sun, J. Li and L. Wang, Phys. Chem. Chem. Phys., 2015, 17, 22072–22078.
33. N. B. Shukla and G. Madras, J. Appl. Polym. Sci., 2012, 126, 463–472.
34. H. Yan, W. X. Zhang, X. W. Kan, L. Dong, Z. W. Jiang, H. J. Li, H. Yang and R. S. Cheng, Colloids Surf., A, 2011, 380, 143–151.
35. M. Auta and B. H. Hameed, J. Ind. Eng. Chem., 2013, 19, 1153–1161.
36. G. J. Copello, A. M. Mebert, M. Raineri, M. P. Pesenti and L. E. Diaz, J. Hazard. Mater., 2011, 186, 932–939.
37. V. K. Gupta, I. Ali, Suhas and D. Mohan, J. Colloid Interface Sci., 2003, 265, 257–264.
38. S. Banerjee, V. Nigam and M. C. Chattopadhyaya, J. Indian Chem. Soc., 2013, 90, 1211–1221.
39. Y. Li, Q. Du, T. Liu, X. Peng, J. Wang, J. Sun, Y. Wang, S. Wu, Z. Wang, Y. Xia and L. Xia, Chem. Eng. Res. Des., 2013, 91, 361–368.
40. W. Shen, S. Chen, S. Shi, X. Li, X. Zhang, W. Hu and H. Wang, Carbohydr. Polym., 2009, 75, 110–114.
41. A. S. Patra, S. Ghorai, S. Ghosh, B. Mandal and S. Pal, J. Hazard. Mater., 2016, 301, 127–136.
42. J. Lutzenkirchen, J. Colloid Interface Sci., 1997, 195, 149–155.
43. Y. Wang, J. Chen, L. Lu and Z. Lin, ACS Appl. Mater. Interfaces, 2013, 5, 7698–7703.
44. L. J. Yang, Y. Y. Zhang, X. Y. Liu, X. Q. Jiang, Z. Z. Zhang, T. T. Zhang and L. Zhang, Chem. Eng. J., 2014, 246, 88–96.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra09049d

---