One-pot synthesis of g-C₃N₄-doped amine-rich porous organic polymer for chlorophenol removal

Abstract

A novel graphitic carbon nitride (g-C₃N₄)/amine-rich porous organic polymer (RAPOP) was synthesized via one-pot polymerization using various molar ratios of melamine (MA)/terephthalaldehyde (TA)/g-C₃N₄ of 4/4/1, 4/4/2 and 4/4/4. The g-C₃N₄ was used as a supporting scaffold, and its stratified structure provided a skeleton. Polymerization between MA and TA mostly occurred on the surface of g-C₃N₄, which formed porous spatial structures, in particular, that of MA/TA/g-C₃N₄ (4/4/2), of which the surface area and pore volume reached 540.36 m² g⁻¹ and 1.502 cm³ g⁻¹, respectively. Their excellent adsorption performance towards 2,4-dichlorophenol was investigated under different conditions. A solution pH of 7 was favorable for adsorption. The presence of Na⁺ and Cl⁻ ions had no adverse effects on the adsorption process, whereas humic acid (HA) led to a slight decrease in performance. Adsorption equilibrium was reached within 40 seconds, and the maximum adsorbed amounts were 188.70 mg g⁻¹, 217.39 mg g⁻¹, 238.10 mg g⁻¹ and 270.27 mg g⁻¹ for MA/TA (4/4), MA/TA/g-C₃N₄ (4/4/4), MA/TA/g-C₃N₄ (4/4/1) and MA/TA/g-C₃N₄ (4/4/2), respectively, at 298 K. Thermodynamic tests indicated that the adsorption proceeded spontaneously and endothermically. Moreover, the adsorbents maintained their high performance and stability after regeneration via treatment with alkali. This work demonstrates that g-C₃N₄/RAPOP can be practically employed to remove chlorophenols from aqueous solutions.

---

Environmental significance

Currently, the discharge of harmful organic pollutants, including phenols, fatty acids, nitriles, etc., from households, agriculture and industry poses a great threat to the environment and human health. Chlorophenols, which are a typical class of phenols, have attracted growing attention because they are employed in many fields, for example, as pesticides and insecticides. They contain a stable conjugated system because of the presence of chlorine atoms and a benzene ring. As a result, they possess the properties of high toxicity, poor biodegradability and biological accumulation. They may induce damage to the structure of cellular phospholipid bilayers, which can cause carcinogenic and mutagenic effects. Therefore, there is an urgent need to rationally and efficiently treat discharged chlorophenol contaminants in wastewater.

1. Introduction

Currently, the discharge of harmful organic pollutants, including phenols, fatty acids, nitriles, etc., from households, agriculture and industry poses a great threat to the environment and human health.¹ Chlorophenols, which are a typical class of phenols, have attracted growing attention because they are employed in many fields, for example, as pesticides and insecticides. They contain a stable conjugated system because of the presence of chlorine atoms and a benzene ring. As a result, they possess the properties of high toxicity, poor biodegradability and biological accumulation.^2,3 They may induce damage to the structure of cellular phospholipid bilayers, which can cause carcinogenic and mutagenic effects.⁴ Therefore, there is an urgent need to rationally and efficiently treat discharged chlorophenol contaminants in wastewater.

Various treatments, such as biocatalysis, adsorption and oxidation, have been established for treating chlorophenols.⁵ Among these, adsorption has mainly attracted attention owing to its high efficiency, low cost and simple implementation. Various traditional adsorbents have been widely employed to remove chlorophenols. For example, Fanaie et al. removed 4-chlorophenol from aqueous solutions by using digested sludge.⁶ Oh et al. studied the use of granular activated carbon to adsorb a binary mixture of phenol and 4-chlorophenol.⁷ Soltani and Lee investigated the adsorption of 2-chlorophenol onto ultrasonically synthesized graphene materials.⁸ However, it can be seen that the adsorption capacity and rate of these materials were severely limited because of defects in their structure. Therefore, the development of novel adsorbent materials with large surface areas and porous structures is urgently required to increase adsorption capacities and accelerate adsorption rates.

In recent years, porous materials have been widely employed, such as covalent organic frameworks (COFs), metal–organic frameworks (MOFs) and polymer-based porous materials.^9–12 Among these, polymer-based porous materials have attracted growing attention because they are endowed with the qualities of low density and high thermal and chemical stability, are linked by strong covalent bonds, and allow pore size control and functional tailoring.^13–15 In particular, RAPOP was prepared by a polymerization process between an amine and phthalaldehyde, exhibited a macro/meso/microporous structure owing to the nano-aggregation effect^16–18 and also possessed many valuable properties, including a large surface area, excellent porous structure, flexible chemical modification and abundant adsorptive sites.^19–21 However, some drawbacks still existed; for example, the homogeneous porosity of its microscopic framework and the excessive aggregation of its nanoparticles made it imperfect as an outstanding porous structure with diversified adsorption performance. Different methods have been utilized to remedy its defects. For example, Zhang et al. synthesized a mesoporous carbon composite derived from RAPOP co-doped with an appropriate weight (4 g) of magnesium nitrate hexahydrate, and the capture capacity for CO₂ increased from 2.27 mmol g⁻¹ to 2.45 mmol g⁻¹ at 298 K and 1 bar.¹⁷

As a covalently bonded polymeric material, g-C₃N₄ has attracted worldwide attention. It has a graphite-like stratified structure and abundant surplus amino groups. In each layer, the p electrons in the p orbitals of the carbon and nitrogen atoms form a complex conjugated system, which makes g-C₃N₄ very chemically stable. In each layer, hydrogen bonding interactions play a major role owing to the presence of nitrogen and oxygen atoms, which can provide abundant active sites and make g-C₃N₄ possess high thermal stability, excellent chemical stability and high reactivity.^22,23 Hence, the doping of g-C₃N₄ into RAPOP was a significant attempt to provide a new pathway to enhance the adsorption performance of the polymer. Cao et al. fabricated a highly water-selective hybrid membrane by incorporating g-C₃N₄ nanosheets into a sodium alginate polymer (3 wt%), and the permeate flux of water was increased from 1653 g m⁻² h⁻¹ to 2469 g m⁻² h⁻¹ at 76 °C and a feed water concentration of 10 wt%.²⁴ Thus, it can be seen that the capacity was greatly increased by doping an auxiliary support into the polymer. In particular, g-C₃N₄ seems to be one of the most promising supports for improving the applicability of these porous materials.

The selective use of g-C₃N₄ as a supporting scaffold plays an important role in facilitating the formation of architectures with large surface areas and outstanding porosity. In the synthesis of RAPOP, there were abundant uncertainties: the particle growth of the polymer was random and occurred at any possible position, and the specific surface area and pore size could easily be affected by internal and external conditions, such as the concentration of the monomers and moisture content of the solution.²⁵ Upon the addition of g-C₃N₄, its particular stratified structure could provide a skeleton for polymerization, and the particle growth of the aggregates could be controlled and channeled in a specific direction. Therefore, the final product was synthesized according to a designed route.^23,26

Here, g-C₃N₄/RAPOP was synthesized by using different molar ratios of MA/TA/g-C₃N₄ of 4/4/1, 4/4/2 and 4/4/4, and RAPOP was prepared via a Schiff base reaction between MA and TA. To evaluate the adsorption ability of samples towards chlorophenols, we selected 2,4-DCP as a model pollutant. Batch adsorption experiments were used to investigate the effects of the solution pH, adsorbent dosage, ionic strength, contact time, temperature and addition of humic substances on the removal of 2,4-DCP. In addition, the kinetics, isotherms and thermodynamics of adsorption were also discussed to study the underlying mechanism. Finally, the reusability and regenerability of the adsorbents were assessed.

2. Materials and methods

2.1. Materials

Melamine (MA) and terephthalaldehyde (TA) were dried at 100 °C in a vacuum. Dimethyl sulfoxide (DMSO) was distilled in a vacuum by drying over CaH₂. 2,4-Dichlorophenol (2,4-DCP) was obtained from Aladdin, and other solvents and reagents were of analytical grade and were purchased from Sinopharm Chemical Reagent Co.

2.2. Preparation of RAPOP and g-C₃N₄/RAPOP

A sample of g-C₃N₄ was prepared via calcination.^27,28 In order to obtain g-C₃N₄ with surplus amine groups, 350 °C was used as the temperature for calcination. In detail, 5 g of MA in an alumina crucible was heated to 350 °C from room temperature in a tube furnace at a rate of 2.5 °C min⁻¹ in an N₂ atmosphere at ambient pressure and kept for 4 h at 350 °C. The synthetic route is displayed in Scheme 1(a). A light yellow sample of g-C₃N₄ was obtained after natural cooling and then ground to form powdered g-C₃N₄ (235 g mol⁻¹).

Scheme 1 Synthetic routes and schematic illustrations of (a) g-C₃N₄, (b) RAPOP and (c) g-C₃N₄/RAPOP.

RAPOP and g-C₃N₄/RAPOP were synthesized by thermal polymerization.^18,29 In order to obtain amine-rich polymers and study the contribution of g-C₃N₄ to the structure and performance of the polymers, the molar ratio of MA/TA was 4/4 and those of MA/TA/g-C₃N₄ were 4/4/1, 4/4/2 and 4/4/4. In detail, MA (0.0100 mol), g-C₃N₄ (0.0000, 0.0025, 0.0050, and 0.0100 mol), TA (0.0100 mol) and DMSO (50 mL) were added to a 100 mL single-necked flask and stirred for 3 hours at 100 °C. Then, 2 mL toluene was added and the mixture was heated for 72 hours at 180 °C under inert conditions using a Dean–Stark apparatus. The synthetic routes are shown in Scheme 1(b and c). After natural cooling, 30 mL acetone was added and the mixture was stirred for 3 hours. Then, acetone, tetrahydrofuran and methylene chloride were separately used to wash the coarse solid. Finally, the product was dried for 12 hours at 100 °C in a vacuum, and the yields reached 65%, 78%, 83% and 72%, respectively.

2.3. Characterization

FE-SEM images were acquired using an FEI Sirion 200 (FEI Co., Netherlands) field emission scanning electron microscope at 10 kV. TEM images were recorded with a Tecnai G220 transmission electron microscope (FEI, Netherlands). Fourier transform infrared (FT-IR) spectra were recorded using a Vertex 70 spectrometer (Bruker, USA). Thermogravimetric analysis (TGA) was performed with a Pyris 1 TGA (PerkinElmer Instruments) thermal analyzer at a rate of 10 °C min⁻¹ under an N₂ atmosphere. X-ray diffraction (XRD) patterns were recorded using a Rigaku D/max-2550 diffractometer with Ni-filtered Cu Kα radiation in the 2θ range from 5° to 80° at 40 kV and 30 mA. X-ray photoelectron spectroscopy (XPS) was performed using a Vacuum Generators MultiLab 2000 spectrometer with monochromatic Al Kα radiation at 298 K. N₂ adsorption–desorption isotherms were recorded with a Micromeritics 2020 HD88 (Micromeritics Instrument Co., USA) apparatus at 77 K. Surface areas were calculated by the Brunauer–Emmett–Teller method. Zeta potentials were measured with a Zetasizer Nano ZS90 analyzer (Malvern Instruments, UK).

2.4. Adsorption experiments

Batch experiments were carried out to determine the effects of the solution pH, adsorbent dosage, ionic strength, contact time, temperature and addition of HA. Before the experiments, the adsorbents were dried at 100 °C for 12 hours in a vacuum. A solution of 2,4-DCP was obtained by dissolving the solid in high-purity water.

The effect of pH was determined by mixing 10 mg samples of adsorbent into 40 mL of 20 mg L⁻¹ 2,4-DCP solutions (pH 2 to 10), which were then shaken in an orbital shaker at 180 rpm and 298 K for 2 minutes to ensure that adsorption equilibrium was reached. The solution pH was adjusted with 0.1 mol L⁻¹ HCl and 0.1 mol L⁻¹ NaOH solutions and measured with a Mettler Toledo FE20 pH meter. Finally, the solutions were separated from the adsorbents with a 0.45 mm syringe filter and analyzed using a Persee TU-1901 UV-vis spectrophotometer at 284 nm.⁵

The adsorbed amount q_t (mg g⁻¹) was calculated as follows:

and the removal rates were calculated by the following equation:where C₀ and C_t (mg L⁻¹) are the concentrations of 2,4-DCP before and after adsorption, respectively, m (mg) is the weight of the adsorbent and V (L) is the volume of the solution.

In order to select the adsorbent dosage, different masses of RAPOP and g-C₃N₄/RAPOP (1, 3, 5, 8, 10, 13, 15, and 20 mg) were added to 2,4-DCP solutions (pH 7, 40 mL, 20 mg L⁻¹) for 2 minutes, respectively. To study the effect of ionic strength, different concentrations of NaCl (2 g L⁻¹ to 10 g L⁻¹) were added to 2,4-DCP solutions. To determine the equilibrium time, the reacted 2,4-DCP solutions (pH 7, 40 mL, 20 mg L⁻¹) were tested at different times. The effect of temperature was determined by investigating the adsorption capacities of the as-prepared materials in 2,4-DCP solutions (pH 7, 40 mL) with different initial concentrations (10, 20, 50, 100, 150, and 200 mg L⁻¹) for 2 minutes at 298 K, 308 K and 318 K, respectively. The effect of the addition of humic substances was investigated by gradually increasing the concentration (30 to 150 mg L⁻¹) of humic acid (HA) in 2,4-DCP solutions, and the HA solution was analyzed using a Teledyne Tekmar TOC analyzer. Finally, to recycle the adsorbed materials, a 0.5 mol L⁻¹ NaOH solution was utilized to elute the adsorbed RAPOP and g-C₃N₄/RAPOP until no 2,4-DCP was detected in the eluate.

3. Results and discussion

3.1. Characterization of RAPOP and g-C₃N₄/RAPOP

FE-SEM and TEM were used to observe their microstructures. As seen in Fig. 1(a and b), g-C₃N₄ possessed a typical layered structure. In each layer, the component particles were arranged closely and regularly. Between neighbouring layers, van der Waals forces were beneficial for forming the layered structure. As shown in Fig. 1(c and d), interconnected nanoparticles with sizes of 20–40 nm formed the skeleton of RAPOP, and a micro/meso/macroporous structure and three-dimensional network framework were both clearly observed.^17,18,29 According to Fig. 1(e–l), the petaloid sheet structure and porous three-dimensional skeleton of g-C₃N₄/RAPOP were perfectly obtained, which suggests that the use of g-C₃N₄ as a supporting scaffold improved the porous structure. The incorporation of g-C₃N₄ provided a skeleton and more abundant reactive sites, and polymerization mostly occurred on the surface of g-C₃N₄, which made the product nanoparticles grow to form agglomerates along every sheet of the stratified structure and finally form more complex porous structures.²⁶ However, when excess g-C₃N₄ was added, the nanoparticle sizes increased and the interstitial spaces became gradually narrower, which consequently resulted in a deterioration of the porous structure.

Fig. 1 FE-SEM and TEM images of (a and b) g-C₃N₄, (c and d) RAPOP and g-C₃N₄/RAPOP with MA/TA/g-Cl₃N₄ ratios of (e–g) 4/4/1, (h–j) 4/4/2 and (k and l) 4/4/4.

The FT-IR spectra are displayed in Fig. 2A. As seen in the spectra, g-C₃N₄ displayed typical absorption bands due to C=N stretching (1572 cm⁻¹), aromatic C–N stretching (1425 cm⁻¹) and deformation modes of s-triazine rings (810 cm⁻¹).^22,30 These indicated that the g-C₃N₄ precursor was successfully prepared. For RAPOP, absorption peaks (Fig. 2A(b)) due to quadrant and semicircle stretching vibrations of triazine rings (1477 cm⁻¹ and 1546 cm⁻¹) and C–N stretching vibrations (1600 cm⁻¹) were observed.^31–33 These results confirmed that RAPOP was polymerized to a significant extent. For g-C₃N₄/RAPOP (Fig. 2A(c–e)), the RAPOP peaks for triazine rings (1477 cm⁻¹ and 1546 cm⁻¹) and C–N stretching vibrations (1600 cm⁻¹) were still retained, and characteristic peaks of g-C₃N₄ could also be found at 1572 cm⁻¹, 1425 cm⁻¹ and 810 cm⁻¹. This demonstrated that g-C₃N₄ had been successfully incorporated into RAPOP. Characteristic peaks of 2,4-DCP (Fig. 2A(f)) were found at 3439 cm⁻¹ and 710 cm⁻¹ for phenolic hydroxyl groups and benzene rings, respectively, and at 1180 cm⁻¹ for chlorine atoms at ortho- and para-positions to phenol groups.^34,35 From Fig. 2A(g), it is obvious that the absorption bands of 2,4-DCP were present in the spectrum of the MA/TA/g-C₃N₄ (4/4/2) polymer, which revealed that 2,4-DCP was adsorbed on the polymer surface.

Fig. 2 (A) FT-IR spectra of (a) g-C₃N₄, (b) RAPOP, g-C₃N₄/RAPOP with MA/TA/g-C₃N₄ ratios of (c) 4/4/1, (d) 4/4/2 and (e) 4/4/4, (f) 2,4-DCP and (g) 2,4-DCP adsorbed on g-C₃N₄/RAPOP with MA/TA/g-C₃N₄ (4/4/2). (B) TGA curves, (C) XRD patterns, and (D) N₂ adsorption–desorption isotherms and (inset) pore size distributions of RAPOP and g-C₃N₄/RAPOP.

TGA was used to investigate the thermostability of the samples (Fig. 2B). RAPOP and g-C₃N₄/RAPOP exhibited slight weight losses below 250 °C owing to the release of residual moisture, solvent and carbon dioxide. Obvious weight loss was not observed until 400 °C, which suggested their outstanding thermal stability. These results were also found for other similar polymers.^17,36 The weight of g-C₃N₄ was reduced dramatically above 350 °C owing to the loss of residual amine groups and a proportion of the nitrogen atoms in triazine rings. Importantly, the weight loss was much less when the molar ratio was increased gradually, which revealed that the incorporation of g-C₃N₄ improved the thermal stability of the polymers.

The XRD patterns of the samples are shown in Fig. 2C. The sharp diffraction peaks of g-C₃N₄ at 12.0°, 14.3°, 26.4° and 29.1° were attributed to the (100), (100), (002) and (102) crystallographic planes, respectively, which implied that the prominent layered structure was in the crystalline state.^22,26,39 The patterns of g-C₃N₄/RAPOP displayed similar peaks to those of g-C₃N₄, which indicated that g-C₃N₄ was successfully incorporated into the polymers and the crystalline phase mainly comprised g-C₃N₄ in these samples of g-C₃N₄/RAPOP. However, with a decrease in the g-C₃N₄ ratio, less intense diffraction peaks of g-C₃N₄/RAPOP were observed, which indicated that RAPOP was in the amorphous state and g-C₃N₄ maintained its layered structure and served the function of supporting the polymer structure.

XPS was performed to analyze the surface binding state and elemental composition. Fig. 3 shows the XPS spectra of MA/TA/g-C₃N₄ (4/4/2). From Fig. 3(a), it is obvious that the polymer contained C, N and O elements, and their mass contents were 59.85%, 32.68% and 7.47%, respectively (inset of Fig. 3(a)), which indicated that the polymer mainly comprised C and N elements. In the high-resolution C 1s scan (Fig. 3(b)), there were three peaks at 287.7, 284.9 and 286.7 eV, which represented carbon atoms in triazine rings (–C=N–), aromatic rings (–C=C–) and –N–C–N– linkages, respectively, which indicated that the monomers were successfully polymerized.^12,22 The high-resolution N 1s scan (Fig. 3(c)) was fitted to four peaks at 398.2, 399.4, 400.0 and 398.8 eV, which corresponded to triazine groups (–C=N–), tertiary amine groups (–N–(C)₃), imine groups (–NH–) and amine groups (–NH₂), respectively.²⁶ The two peaks for O 1s at 533.4 and 532.1 eV (Fig. 3(d)) resulted from a small quantity of carbonyl groups (–C=O) and hydroxyl groups (–OH), respectively, which belonged to surface adsorbed carbon dioxide and moisture.¹⁷ These results all demonstrated that MA and TA were polymerized to a sufficient extent and g-C₃N₄ was successfully incorporated into the polymers to form the novel g-C₃N₄/RAPOP polymer.

Fig. 3 XPS spectra of the polymer (MA/TA/g-C₃N₄ (4/4/2)): (a) survey scan and high-resolution scans of (b) C 1s, (c) N 1s, and (d) O 1s, and the mass contents (inset of (a)).

N₂ adsorption–desorption isotherms (Fig. 2D) were employed to analyze the pore properties of the materials at 77 K, and the parameters are listed in Table 1. All the isotherms were type II isotherms with an H3 hysteresis loop, which indicated the aggregation of nanoparticles containing slit-shaped pores. At P/P₀ < 0.1, a sharp upward trend in nitrogen adsorption appears, indicating the existence of a microporous structure. At P/P₀ > 0.9, the rapid increase in the isotherm indicated the presence of macropores, which originated from interspaces between interconnected nanoparticles.³⁷ In comparison with RAPOP, which had a Brunauer–Emmett–Teller (BET) surface area of 368.05 m² g⁻¹ and a total pore volume of 0.651 cm³ g⁻¹, as calculated by non-local density functional theory (NLDFT), g-C₃N₄/RAPOP displayed excellent ability and high particle consistency. In detail, BET surface areas of 489.73, 540.36 and 445.19 m² g⁻¹ and total pore volumes of 1.175, 1.502 and 0.936 cm³ g⁻¹ were measured for MA/TA/g-C₃N₄ (4/4/1), MA/TA/g-C₃N₄ (4/4/2) and MA/TA/g-C₃N₄ (4/4/4), respectively. Presumably, the g-C₃N₄ dopant introduced unreacted amine groups, which could also react with the aldehyde groups of TA and form hydrogen bonds between the amine-containing polymers, which contributed to the formation of a more complex porous structure.^29,38 In terms of the pore size distribution (inset of Fig. 2D), there was a significant difference between RAPOP and g-C₃N₄/RAPOP. RAPOP predominantly contained micro- and mesopores with a Barrett–Joyner–Halenda (BJH) average pore size of 4.23 nm. However, the macropores in g-C₃N₄/RAPOP gradually increased in proportion upon doping with g-C₃N₄, and the BJH average pore sizes were 5.68, 6.33 and 5.19 nm for MA/TA/g-C₃N₄ (4/4/1), MA/TA/g-C₃N₄ (4/4/2) and MA/TA/g-C₃N₄ (4/4/4), respectively. These proved again that the use of g-C₃N₄ as a supporting skeleton was able to greatly improve the architecture with a larger specific surface area and more outstanding porous structure (as illustrated in Scheme 1(c)), which was consistent with the FE-SEM and TEM results shown in Fig. 1.

Table 1 Specific surface areas and pore volume parameters of RAPOP and g-C₃N₄/RAPOP

Materials	S BET /m^2 g^−1	S mic /m^2 g^−1	V mic /cm^3 g^−1	V meso /cm^3 g^−1	V t /cm^3 g^−1	D p /nm
a Determined by N2 adsorption using the Brunauer–Emmett–Teller (BET) method. b Micropore area, determined by DFT. c Micropore volume, calculated using the Dubinin–Astakhov method. d Mesopore volume, calculated as Vt − Vmic. e Total pore volume, determined at P/P0 = 0.9923. f Adsorption average pore width (4V/A by BET).
MA/TA (4/4)	368.05	160.92	0.084	0.567	0.651	4.23
MA/TA/g-C3N4 (4/4/1)	489.73	271.42	0.148	1.027	1.175	5.68
MA/TA/g-C3N4 (4/4/2)	540.36	328.88	0.171	1.331	1.502	6.33
MA/TA/g-C3N4 (4/4/4)	445.19	229.08	0.129	0.807	0.936	5.19

---

3.2. Adsorption performance evaluation under different conditions

3.2.1. Solution pH. The solution pH is a crucial factor that influences the adsorption of pollutants from aqueous solutions, because different pH values affect the surface charge of adsorbent materials and the dissociation properties of organic contaminants, which consequently alters the adsorption capacity. The effect of the pH on the removal of 2,4-DCP is displayed in Fig. 4(a). The q_e values for g-C₃N₄/RAPOP were higher than that for RAPOP, and the MA/TA/g-C₃N₄ (4/4/2) polymer displayed the highest q_e value at all pH values. The results implied that the doping of g-C₃N₄ into RAPOP markedly increased its adsorption capacity, whereas the q_e values for the polymer decreased gradually on the addition of excess g-C₃N₄. The graphite-like stratified structure and surplus amino groups of g-C₃N₄ made a great contribution to the results. When g-C₃N₄ was doped into RAPOP, its adsorption capacity was greatly increased because the doped g-C₃N₄ could incorporate abundant primary amine groups and adsorptive sites, which led to the formation of stronger interaction forces between the polymers and 2,4-DCP. However, when g-C₃N₄ was doped in excess, the adsorption capacity was decreased, which was partly because the porous structure of the polymers was destroyed to some extent, which led to a massive reduction in the number of adsorptive sites.^22,39 Therefore, an MA/TA/g-C₃N₄ molar ratio of 4/4/2 was the optimal ratio for the preparation of g-C₃N₄/RAPOP.

Fig. 4 Effects of (a) solution pH, (b) ionic strength, (c) contact time and (d) humic substances on the adsorption of 2,4-DCP by RAPOP and g-C₃N₄/RAPOP (dosage: 10 mg; solution: 20 mg L⁻¹, 40 mL; temperature: 298 K) and (inset of a) zeta potentials of the polymer (MA/TA/g-C₃N₄ (4/4/2)) at different pH values.

As shown in Fig. 4(a), the polymers exhibited relatively high adsorption capacities at a pH in the range of 2 to 7, whereas these were reduced sharply when the pH exceeded 7. To understand the influence of pH on the removal of 2,4-DCP, the zeta potentials of the MA/TA/g-C₃N₄ (4/4/2) polymer at different pH values were determined (inset of Fig. 4(a)). The isoelectric point was found to be around 5.8, which is consistent with the pK_a value of amine groups in similar polymers.⁴⁰ This means that the surface of the material is positively charged when the pH is below 5.8. Given that 2,4-DCP is a weak acid (pK_a = 7.9),⁴¹ the adsorbed 2,4-DCP mostly existed in the non-ionized form in acidic solutions. Therefore, electrostatic repulsion would not occur between 2,4-DCP and the adsorbent. When the pH values exceeded 7, 2,4-DCP dissociated in basic solutions to form chlorophenolate anions. Moreover, the surface functional groups of the adsorbent were also negatively charged. As a result, the electrostatic repulsion between identical charges reduced the adsorption capacity. Therefore, a pH of 7 is the optimal value and was adopted in the following experiments. Under these conditions, electrostatic interactions could be neglected, and π–π interactions and hydrogen bonding interactions were mainly responsible for the adsorption of 2,4-DCP on the adsorbent surface (see Scheme 2). The backbone structures of RAPOP and g-C₃N₄/RAPOP comprise aromatic rings, and the p orbitals of the carbon and nitrogen atoms in these aromatic rings form a stable conjugated system. 2,4-DCP also contains a conjugated system. Moreover, the hydroxyl groups (–OH) of 2,4-DCP and amine groups (–NH₂) of the adsorbent were abundant. When 2,4-DCP approached the surface of the adsorbent, hydrogen bonding interactions were gradually established between hydroxyl and amine groups, which bound the 2,4-DCP molecules located on the surface of the adsorbent. Furthermore, 2,4-DCP molecules, which contain π electrons, interacted with aromatic rings on the adsorbent surface via π–π interactions, which caused the 2,4-DCP molecules to be tightly arranged on the surface of the adsorbent. Thus, the underlying mechanism is believed to arise from the intrinsically high adsorption ability of the porous organic polymers, as well as the strong interactions between the polymers and 2,4-DCP, including π–π interactions and hydrogen bonding interactions.

Scheme 2 Mechanistic illustration of the adsorption of 2,4-DCP on g-C₃N₄/RAPOP.

3.2.2. Adsorbent dosage. The effect of the adsorbent dosage is shown in Fig. S1.† With an increase in the adsorbent dosage, the removal rate increased markedly, while the q_e value was reduced. This observation could be explained by adsorptive sites and surface energy. A higher adsorbent dosage resulted in an increase in the probability of collisions of adsorbent particles in aqueous solution, which reduced the number of adsorptive sites in the adsorbent and finally decreased the adsorption capacity owing to a reduction in surface energy. However, the removal rate increased because the apparent adsorption capacity of the adsorbent had been enhanced. Furthermore, we also found that the q_e values for the four adsorbents varied as follows: MA/TA/g-C₃N₄ (4/4/2) > MA/TA/g-C₃N₄ (4/4/1) > MA/TA/g-C₃N₄ (4/4/4) > MA/TA (4/4), which indicated that g-C₃N₄ made a remarkable contribution to the adsorption capacity of g-C₃N₄/RAPOP.

3.2.3. Ionic strength. The ionic strength is a non-ignorable factor in aqueous solutions, because ions with different charges can participate in the adsorption process and affect the surface functional groups of the adsorbent. NaCl was selected as a representative to study the effect of the ionic strength on the adsorption process, because Na⁺ and Cl⁻ ions are the most common ions that are present in wastewater that contains chlorophenols. From Fig. 4(b), it is clear that no obvious change was found on increasing the NaCl concentration gradually. The pH-dependent and ion-independent nature of the adsorption process suggested that covalent bonding interactions could be ignored, whereas hydrogen bonding interactions, π–π interactions and electrostatic interactions were responsible for the removal of 2,4-DCP. Thus, ions could not compete with 2,4-DCP adsorbed onto the adsorbent. This also indicated that g-C₃N₄/RAPOP can be employed to remove chlorophenols from wastewater that contains salt.

3.2.4. Contact time. The equilibrium time is a significant parameter that influences practical applications in wastewater treatment systems. The effect of the contact time is shown in Fig. 4(c). It can be seen that the adsorption of 2,4-DCP was rapid for all four adsorbents. Obviously, the reaction was fast for the initial 20 seconds, and saturation was reached at 40 seconds. These results could be explained by the functional adsorptive sites. At first, adsorption proceeded quickly because there were plentiful unoccupied adsorptive sites. While adsorption was occurring, the adsorption process gradually slowed because the residual adsorptive sites were occupied by adsorbed 2,4-DCP. The 2,4-DCP molecules gradually diffused and were adsorbed onto the interior surfaces of the pores.⁴² Equilibrium was reached when the adsorptive sites were saturated. Moreover, the order of the apparent adsorption rates was MA/TA/g-C₃N₄ (4/4/2) > MA/TA/g-C₃N₄ (4/4/1) > MA/TA/g-C₃N₄ (4/4/4) > MA/TA (4/4). This observation implied that the doping of g-C₃N₄ into RAPOP accelerated the adsorption process in comparison with RAPOP.

3.2.5. Temperature. Another critical factor that greatly affects the adsorption process is the system temperature. In general, higher temperatures can accelerate the diffusion of adsorbate particles, reduce the surface activation energy in the adsorption process of the adsorbate, and finally change the saturation adsorption capacity of the adsorbent material.⁴³ The effect of temperature is plotted in Fig. 5. As the temperature rose, the q_e value increased because of a reduction in the residual 2,4-DCP concentration. In particular, at a concentration of 200 mg L⁻¹ the q_e value for the MA/TA/g-C₃N₄ (4/4/2) polymer increased from 270.16 mg g⁻¹ to 281.48 mg g⁻¹ and 295.63 mg g⁻¹ upon an increase in temperature from 298 K to 308 K and 318 K. This implied that the adsorption reaction was endothermic and was favored at higher temperatures.

Fig. 5 Effect of temperature on the adsorption of 2,4-DCP by RAPOP and g-C₃N₄/RAPOP using (a) MA/TA (4/4), (b) MA/TA/g-C₃N₄ (4/4/1), (c) MA/TA/g-C₃N₄ (4/4/2) and (d) MA/TA/g-C₃N₄ (4/4/4) (dosage: 10 mg; solution: 20 mg L⁻¹, 40 mL; pH = 7; time: 2 minutes).

3.2.6. Effects of humic substances on 2,4-DCP removal. Humic acid (HA), which is a typical humic substance, is a common organic metabolite and forms a large proportion of discharged sewage. It may participate in adsorption and compete with adsorbate molecules, which consequently reduces the adsorption capacity. The effect of the addition of HA is shown in Fig. 4(d). For MA/TA (4/4), MA/TA/g-C₃N₄ (4/4/1), MA/TA/g-C₃N₄ (4/4/2) and MA/TA/g-C₃N₄ (4/4/4), a limited decrease in q_e values was found. The q_e values for the virgin solutions equaled 63.99, 76.42, 78.31 and 71.07 mg g⁻¹, respectively. As the HA concentration increased from 0.0 to 150.0 mg L⁻¹, the q_e value initially decreased sharply and finally reached equilibrium at 52.99, 66.98, 70.44 and 62.27 mg g⁻¹, respectively. HA could interact with the adsorbents by π–π interactions, hydrogen bonding, and hydrophobic and electrostatic interactions, because HA always contains aromatic rings and oxygen-containing functional groups (–COOH and Ar–OH).^44,45 At the beginning of the adsorption of 2,4-DCP, many adsorptive sites on the surface of these adsorbents were activated and unoccupied, and negatively charged HA ions would compete with 2,4-DCP molecules to occupy adsorptive sites. Besides, a proportion of free 2,4-DCP molecules might be bound by HA ions via complexation interactions.

3.3. Theoretical discussion of adsorption

3.3.1. Adsorption kinetics. Kinetics studies provide major indicators that determine large-scale applications of processes and reflect their feasibility and efficiency. Pseudo-first-order and pseudo-second-order models were employed to investigate the kinetic performance and study the basic mechanism of adsorption. Nonlinear forms of these models are given in the following equations:⁴⁶

q_t = q_e(1 − e^−k₁t) (3)

where q_e (mg g⁻¹) is the equilibrium adsorbed amount, q_t (mg g⁻¹) is the adsorbed amount at time t, t (s) is the contact time, and k₁ (s⁻¹) and k₂ (g mg⁻¹ s⁻¹) are the rate constants of the pseudo-first-order and pseudo-second-order models, respectively. The model fittings and values of the parameters q_e, k, and R² are shown in Fig. S2† and Table 2.

Table 2 Parameters fitted by pseudo-first-order and pseudo-second-order models

Materials	q e(exp) (mg g^−1)	Pseudo-first-order kinetic model			Pseudo-second-order kinetic model
		k 1 (s^−1)	q e1(cal) (mg g^−1)	R ^2	k 2 (× 10^−3) (g mg^−1 s^−1)	q e2(cal) (mg g^−1)	R ^2
MA/TA (4/4)	63.99	0.177	59.72	0.9336	4.4	65.23	0.9792
MA/TA/g-C3N4 (4/4/1)	76.42	0.160	71.61	0.8940	3.1	78.91	0.9859
MA/TA/g-C3N4 (4/4/2)	78.32	0.174	73.45	0.8853	3.5	80.36	0.9784
MA/TA/g-C3N4 (4/4/4)	71.07	0.164	65.82	0.9052	3.6	72.40	0.9890

---

It was shown that the pseudo-second-order model gave higher R² values for all adsorbents. On the basis of the experimental data (q_e(exp)), the equilibrium adsorbed amounts (q_e2(cal)) obtained from the pseudo-second-order model were more acceptable than those obtained from the pseudo-first-order model, which implied that the pseudo-second-order model could be used to describe the adsorption of 2,4-DCP on RAPOP and g-C₃N₄/RAPOP and revealed that the reaction rates were proportional to the square of the number of free sites.

3.3.2. Adsorption isotherm. The Langmuir model is based on an ideal situation comprising absolutely homogeneous surfaces of adsorbents, monolayer coverage, and identical adsorptive sites. The Freundlich model makes an empirical assumption of heterogeneous adsorption energies and is not restricted to monolayer coverage. Linear forms of these models are given as follows:⁴⁷where q_e (mg g⁻¹) is the equilibrium adsorption capacity, C_e (mg L⁻¹) is the equilibrium adsorbed concentration, q_m (mg g⁻¹) is the maximum adsorption capacity, which corresponds to complete monolayer coverage, b (L mg⁻¹) is the Langmuir constant, which reflects the adsorption energy and capacity, K_f (mg^1−1/n L^1/n g⁻¹) and n are the Freundlich constants, which represent the adsorption capacity and intensity, respectively. The isotherm fittings and values of the parameters q_m, b, n, K_f and R² are shown in Fig. S3† and Table 3.

Table 3 Parameters fitted by Langmuir and Freundlich models

Materials	T (K)	Langmuir model			Freundlich model
		b (L mg^−1)	q max (mg g^−1)	R ^2	n	K f (mg^1−1/n L^1/n g^−1)	R ^2
MA/TA (4/4)	298	0.13315	188.70	0.9990	3.02389	39.6107	0.9622
	308	0.13545	200.00	0.9978	3.12891	43.3454	0.9712
	318	0.13905	212.77	0.9952	3.51741	52.9633	0.9778
MA/TA/g-C3N4 (4/4/1)	298	0.16601	238.10	0.9910	3.74532	66.5597	0.9931
	308	0.16949	250.00	0.9880	4.06835	75.2338	0.9956
	318	0.18095	263.16	0.9850	4.11523	80.9312	0.9934
MA/TA/g-C3N4 (4/4/2)	298	0.15102	270.27	0.9853	3.59195	70.2809	0.9942
	308	0.16216	277.78	0.9821	3.92927	80.3024	0.9970
	318	0.18041	285.71	0.9791	4.18760	90.4503	0.9695
MA/TA/g-C3N4 (4/4/4)	298	0.15282	217.39	0.9958	3.46260	54.7731	0.9810
	308	0.15548	227.27	0.9935	3.56379	59.2165	0.9847
	318	0.16867	238.10	0.9908	4.09333	72.1466	0.9852

---

It was demonstrated that the adsorption could be well described by the Langmuir model according to the values of R², which implied that the adsorptive sites were effectively equivalent to monolayer adsorption. At room temperature, the q_m values for the four adsorbents were 188.70, 217.39, 238.10 and 270.27 mg g⁻¹ for MA/TA (4/4), MA/TA/g-C₃N₄ (4/4/4), MA/TA/g-C₃N₄ (4/4/1) and MA/TA/g-C₃N₄ (4/4/2), respectively. These results indicated that doping the appropriate dosage of g-C₃N₄ into RAPOP could boost its capacity to adsorb 2,4-DCP. Moreover, the values of n were all much greater than 1.0, which implied that this process is favorable for removing 2,4-DCP from wastewater.

3.3.3. Adsorption thermodynamics. A thermodynamic study was carried out because it was seen to provide key indicators of whether the adsorption process can occur spontaneously in natural conditions and to determine the heat change in the adsorption reaction. The change in Gibbs free energy is a basic standard used to decide the spontaneity of a reaction.⁴⁸ The apparent equilibrium constant (K₀) was calculated by the following equation:where C_ad,e (mg) is the equilibrium adsorbed amount of 2,4-DCP per liter of the solution, and C_e (mg L⁻¹) is the equilibrium concentration. The value of K₀ can be obtained at the lowest experimental concentration of 2,4-DCP and can be used to determine the values of ΔG, ΔH and ΔS by the following equations:

ΔG = −RT ln K₀ (8)

ΔG = ΔH − TΔS (9)

where ΔG (kJ mol⁻¹) is the change in Gibbs free energy, ΔH (kJ mol⁻¹) is the change in enthalpy, and ΔS (J mol⁻¹ K⁻¹) is the change in entropy.

The values of the parameters ΔH, ΔS, ΔG and the correlation coefficient (R²) are listed in Table 4. For MA/TA (4/4), MA/TA/g-C₃N₄ (4/4/1), MA/TA/g-C₃N₄ (4/4/2) and MA/TA/g-C₃N₄ (4/4/4) the ΔH values were 1.704, 3.374, 6.987 and 3.859 kJ mol⁻¹, respectively, which indicated that all these adsorption processes were endothermic and were favored at higher temperatures. The ΔS values were 46.346, 53.762, 65.107 and 54.672 J mol⁻¹ K⁻¹, respectively, and the ΔG values at 298, 308 and 318 K were calculated to be −12.107, −12.571, and −13.034 kJ mol⁻¹, −12.647, −13.185, and −13.722 kJ mol⁻¹, −12.415, −13.066, and −13.717 kJ mol⁻¹, and −12.433, −12.980, and −13.527 kJ mol⁻¹, respectively. It was demonstrated that the adsorbent particles had a high affinity towards 2,4-DCP, and the adsorption process occurred spontaneously in natural conditions.

Table 4 Thermodynamic parameters of the adsorption process

Material	T (K)	ΔS (J mol^−1 K^−1)	ΔH (kJ mol^−1)	ΔG (kJ mol^−1)
MA/TA (4/4)	298	46.346	1.704	−12.107
	308			−12.571
	318			−13.034
MA/TA/g-C3N4 (4/4/1)	298	53.762	3.374	−12.647
	308			−13.185
	318			−13.722
MA/TA/g-C3N4 (4/4/2)	298	65.107	6.987	−12.415
	308			−13.066
	318			−13.717
MA/TA/g-C3N4 (4/4/4)	298	54.672	3.859	−12.433
	308			−12.980
	318			−13.527

---

3.4. Recycling and reuse of g-C₃N₄/RAPOP

The reusability of an adsorbent material is a key factor for improving the process economics. The adsorbents consistently retained their porous structure because the adsorbed 2,4-DCP only occupied adsorptive sites on the adsorbent surface, which suggested that the used adsorbents can be easily recycled via common methods, such as desorption or biocatalysis.^49,50 As described for the effect of the solution pH (Fig. 4(a)) in alkaline environments, both the adsorbents and 2,4-DCP were negatively charged and there was strong electrostatic repulsion between them, which led to the desorption of 2,4-DCP from the adsorbents. This observation revealed that the adsorbents can be effectively regenerated via treatment with alkali.

Adsorption–desorption cycles were carried out five times, and the results are displayed in Fig. 6. For MA/TA (4/4), MA/TA/g-C₃N₄ (4/4/1), MA/TA/g-C₃N₄ (4/4/2) and MA/TA/g-C₃N₄ (4/4/4), the q_e values for the virgin adsorbents equaled 63.99, 76.42, 78.31 and 71.07 mg g⁻¹, respectively, whereas the regenerated adsorbents displayed outstanding regeneration with a slight reduction in the value of q_e. Specifically, q_e values of 57.24, 69.82, 72.32 and 64.53 mg g⁻¹, respectively, were measured after five recycling runs. From the FT-IR spectra and TEM images of regenerated RAPOP and g-C₃N₄/RAPOP after five recycling runs (see Fig. S4 and S5†), the characteristic peaks of triazine rings (1477 cm⁻¹ and 1546 cm⁻¹), C–N stretching vibrations (1600 cm⁻¹) and g-C₃N₄ (1572 cm⁻¹, 1425 cm⁻¹ and 810 cm⁻¹) were retained, which implied that the regenerated polymers retained their functional groups and chemical bonds. Importantly, the porous structures with interconnected nanoparticles were completely preserved. These results all indicated that g-C₃N₄/RAPOP possessed excellent regenerability. The highly abundant pores, interconnecting pore channels and large specific surface areas of the as-prepared materials could greatly reduce the possibility of pore blockage by adsorbate molecules. The alkaline environment also made the adsorbents achieve excellent recycling performance. The outstanding regeneration of g-C₃N₄/RAPOP demonstrated good potential for the recovery and abatement of phenolic compounds in wastewater treatment.

Fig. 6 Effect of successive recycling runs on the adsorption of 2,4-DCP by RAPOP and g-C₃N₄/RAPOP.

3.5. Comparison with other adsorbents

To position the novel adsorbents with respect to other reported adsorbents that have also been used to remove 2,4-DCP, a comparison was made between the results of the present work and reported data. According to the results of the Langmuir model (see Table 3), the maximum adsorption capacities (q_max) reached 188.70, 238.10, 270.27 and 217.39 mg g⁻¹ for MA/TA (4/4), MA/TA/g-C₃N₄ (4/4/1), MA/TA/g-C₃N₄ (4/4/2) and MA/TA/g-C₃N₄ (4/4/4), respectively, at 298 K. The adsorption capacities of other reported adsorbent materials for 2,4-DCP are listed in Table 5, which shows that RAPOP and g-C₃N₄/RAPOP exhibited excellent performance in the removal of the organic contaminant. In particular, the MA/TA/g-C₃N₄ (4/4/2) polymer had a higher adsorption capacity than the other adsorbents, which implied the efficient adsorption properties of g-C₃N₄/RAPOP towards 2,4-DCP. Hydrogen bonding interactions, π–π interactions and pore filling might be responsible for the higher adsorption capacity. Therefore, it can be positively concluded that g-C₃N₄/RAPOP has great potential for the adsorption and recovery of chlorophenols from wastewater.

Table 5 Comparison of adsorption capacity for 2,4-DCP of different adsorbents

Adsorbents	T (K)	pH	q max (mg g^−1)	Ref.
Marine sediment	298	7.0	3.85	51
Rice husk	298	—	40.50	52
Cattail fiber-based activated carbon	293	—	142.85	53
Coir pith activated carbon	308	7.0	5.78	54
Oil palm empty fruit bunch carbon	303	6.0	27.25	55
Maize cob carbon	303	—	17.94	56
Multiwalled carbon nanotube	298	6.0	138.00	57
Palm pith carbon	298	7.0	19.16	58
β-CD/attapulgite composites	308	7.0	19.04	59
Cyclodextrin polymers	298	7.0	15.70	60
MA/TA (4/4)	298	7.0	188.70	This work
MA/TA/g-C3N4 (4/4/1)	298	7.0	238.10	This work
MA/TA/g-C3N4 (4/4/2)	298	7.0	270.27	This work
MA/TA/g-C3N4 (4/4/4)	298	7.0	217.39	This work

---

4. Conclusions

In summary, g-C₃N₄/RAPOP was successfully synthesized by using different molar ratios of MA/TA/g-C₃N₄ of 4/4/1, 4/4/2 and 4/4/4, and RAPOP was prepared by a Schiff base reaction between MA and TA. The use of g-C₃N₄ as a supporting skeleton in RAPOP produced three major beneficial effects: (a) the specific surface areas and total pore volumes increased and were highest for MA/TA/g-C₃N₄ (4/4/2), for which they reached 540.36 m² g⁻¹ and 1.502 cm³ g⁻¹, respectively. (b) The larger specific surface area, more outstanding porous structure, and surplus amine groups combined to improve the adsorption performance towards 2,4-DCP. The equilibrium time and highest maximum adsorption capacity were 40 seconds and 270.27 mg g⁻¹ at 298 K, respectively. (c) The adsorption process occurred spontaneously, and the polymers exhibited excellent regeneration after treatment with alkali. Consequently, we fully believe that such a highly efficient, easily synthesized, and low-cost adsorbent can be employed in the treatment of wastewater and drinking water contaminated by phenolic compounds.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Scientific Foundation of China (21477118, 51478445, 51678546).

References

1. O. S. Sanyal, Z. G. Liu, J. Yu, B. M. Meharg, J. S. Hong, W. Liao and I. Lee, J. Membr. Sci., 2016, 512, 21–28.
2. Y. Watanabe, T. Harada, H. Kawai, T. Kaji, M. Hiramoto and K. Nishiyama, J. Mol. Liq., 2016, 217, 51–56.
3. N. Tandjaoui, M. Abouseoud, A. Couvert, A. Amrane and A. Tassist, Chemosphere, 2016, 148, 55–60.
4. T. T. Li, G. Z. Lu, X. L. Hu, W. Q. Xie, Q. Xia and S. L. Luo, Mater. Lett., 2017, 188, 392–395.
5. D. Panda and S. Manickam, Ultrason. Sonochem., 2017, 36, 481–496.
6. V. R. Fanaie, M. Karrabi, M. M. Amin, B. Shahnavaz and A. Fatehizadeh, Int. J. Environ. Sci. Technol., 2017, 14, 37–48.
7. W. D. Oh, P. E. Lim, K. Y. Leong, S. L. Yong and H. Yin, Desalin. Water Treat., 2016, 57, 20476–20482.
8. T. Soltani and B. K. Lee, J. Colloid Interface Sci., 2016, 481, 168–180.
9. Y. T. Xia, Y. K. Li, Y. T. Gu, T. Jin, Q. Yang, J. Hu, H. L. Liu and H. L. Wang, Fuel, 2016, 170, 100–106.
10. J. Y. Zhang, D. Xu, W. J. Qian, J. Y. Zhu and F. Yan, Carbon, 2016, 105, 183–190.
11. J. B. Lee, M. K. Ahn, Y. H. Koh, H. Lee and H. E. Kim, Mater. Des., 2016, 101, 323–331.
12. G. W. Yang, H. Y. Han, C. Y. Du, Z. H. Luo and Y. J. Wang, Polymer, 2010, 51, 6193–6202.
13. M. Ghani, R. M. Frizzarin, F. Maya and V. Cerda, J. Chromatogr. A, 2016, 1453, 1–9.
14. R. E. Osta, A. Carlin-Sinclair, N. Guillou, R. I. Walton, F. Vermoortele, M. Maes, D. D. Vos and F. Millange, Chem. Mater., 2012, 24, 2781–2791.
15. M. W. Dunlop, P. J. Blackall and R. M. Stuetz, J. Environ. Manage., 2016, 177, 306–319.
16. Y. L. Luo, B. Y. Li, L. Y. Liang and B. Tan, Chem. Commun., 2011, 47, 7704–7706.
17. Z. Z. Zhang, C. M. Zhu, N. N. Sun, H. Wang, Z. Y. Tang, W. Wei and Y. H. Sun, J. Phys. Chem. C, 2015, 119, 9302–9310.
18. H. J. Ou, Q. L. You, J. Li, G. Y. Liao, H. Xia and D. S. Wang, RSC Adv., 2016, 6, 98487–98497.
19. F. B. Su, C. K. Poh, J. S. Chen, G. W. Xu, D. Wang, Q. Li, J. Y. Lin and X. W. Lou, Energy Environ. Sci., 2011, 4, 717–724.
20. Y. Z. Liao, J. Weber and C. F. J. Faul, Macromolecules, 2015, 48, 2064–2073.
21. R. M. Li, A. M. Cao, Y. J. Zhang, G. Li, F. Jiang, S. M. Li, D. Q. Chen, C. R. Wang, J. C. Ge and C. Y. Shu, ACS Appl. Mater. Interfaces, 2014, 6, 20574–20578.
22. Y. X. Zeng, C. B. Liu, L. L. Wang, S. Q. Zhang, Y. B. Ding, Y. Z. Xu, Y. T. Liu and S. L. Luo, J. Mater. Chem. A, 2016, 4, 19003–19010.
23. M. Wu, J. M. Yan, X. W. Zhang and M. Zhao, Appl. Surf. Sci., 2015, 354, 196–200.
24. K. T. Cao, Z. Y. Jiang, X. S. Zhang, Y. M. Zhang, J. Zhao, R. S. Xing, S. Yang, C. Y. Gao and F. S. Pan, J. Membr. Sci., 2015, 490, 72–83.
25. X. L. Zhao and N. Yan, RSC Adv., 2015, 5, 69955–69961.
26. H. H. Liu, D. L. Chen, Z. Q. Wang, H. J. Jing and R. Zhang, Appl. Catal., B, 2017, 203, 300–313.
27. Y. J. Zhong, Z. Q. Wang, J. Y. Feng, S. C. Yan, H. T. Zhang, Z. S. Li and Z. G. Zou, Appl. Surf. Sci., 2014, 295, 253–259.
28. P. Martin-Ramos, J. Martin-Gil, R. C. Dante, F. Vaquero, R. M. Navarro and J. L. G. Fierro, Int. J. Hydrogen Energy, 2015, 40, 7273–7281.
29. J. X. Hu, H. Shang, J. G. Wang, L. Luo, Q. Xiao, Y. J. Zhong and W. D. Zhu, Ind. Eng. Chem. Res., 2014, 53, 11828–11837.
30. W. N. Xing, C. M. Li, G. Chen, Z. H. Han, Y. S. Zhou, Y. D. Hu and Q. Q. Meng, Appl. Catal., B, 2017, 203, 65–71.
31. S. Uysal, J. Inclusion Phenom. Macrocyclic Chem., 2013, 76, 223–230.
32. S. Uysal and A. N. Kursunlu, J. Inorg. Organomet. Polym., 2011, 21, 291–296.
33. X. Q. Lin, X. Z. Li, F. Li, Y. Y. Fang, M. Tian, X. C. An, Y. Fu, J. Jin and J. T. Ma, J. Mater. Chem. A, 2016, 4, 6505–6512.
34. Z. P. Cao, M. H. Zhang, J. L. Zhang and H. W. Zhang, Bioresour. Technol., 2016, 212, 138–143.
35. Y. J. Qian, J. Zhang, Y. L. Zhang, J. B. Chen and X. F. Zhou, Sep. Purif. Technol., 2016, 166, 222–229.
36. D. H. Yang, Z. Q. Yao, D. H. Wu, Y. H. Zhang, Z. Zhou and X. H. Bu, J. Mater. Chem. A, 2016, 4, 18621–18627.
37. J. S. Wei, H. Ding, Y. G. Wang and H. M. Xiong, ACS Appl. Mater. Interfaces, 2015, 7, 5811–5819.
38. J. H. Kou and L. B. Sun, Ind. Eng. Chem. Res., 2016, 55, 10916–10925.
39. J. J. Wang, L. Tang, G. M. Zeng, Y. N. Liu, Y. Y. Zhou, Y. C. Deng, J. J. Wang and B. Peng, ACS Sustainable Chem. Eng., 2017, 5, 1062–1072.
40. S. Fujii, M. Kido, M. Sato, Y. Higaki, T. Hirai, N. Ohta, K. Kojioa and A. Takahara, Polym. Chem., 2015, 6, 7053–7059.
41. H. Yuan, Q. L. You, L. J. Song, G. Y. Liao, H. Xia and D. S. Wang, RSC Adv., 2016, 6, 95825–95835.
42. M. Y. Guo, X. L. Weng, T. Wang and Z. L. Chen, Sep. Purif. Technol., 2017, 175, 222–228.
43. A. Demirak, O. Dalman, E. Tilkan, D. Yildiz, E. Yavuz and C. Gokce, Microchem. J., 2011, 99, 97–102.
44. F. Wang, S. Ma, Y. Si, L. F. Dong, X. L. Wang, J. Yao, H. L. Chen, Z. J. Yi, W. C. Yao and B. S. Xing, Carbon, 2017, 114, 671–678.
45. Y. L. Kong, J. Kang, J. M. Shen, Z. L. Chen and L. T. Fan, Environ. Sci. Pollut. Res., 2017, 24, 2381–2393.
46. P. Tian, X. Y. Han, G. L. Ning, H. X. Fang, J. W. Ye, W. T. Gong and Y. Lin, ACS Appl. Mater. Interfaces, 2013, 5, 12411–12418.
47. M. V. Subbaiah and D. S. Kim, Ecotoxicol. Environ. Saf., 2016, 128, 109–117.
48. Q. Liu, L. B. Zhong, Q. B. Zhao, C. Frear and Y. M. Zheng, ACS Appl. Mater. Interfaces, 2015, 7, 14573–14583.
49. A. Meng, J. Xing, Z. J. Li and Q. D. Li, ACS Appl. Mater. Interfaces, 2015, 7, 27449–27457.
50. Y. J. Zhang, B. Prigent and S. U. Geißen, Chemosphere, 2016, 154, 132–137.
51. P. Wu, G. P. Yang and X. K. Zhao, J. Colloid Interface Sci., 2003, 265, 251–256.
52. V. Vadivelan and K. Kumar, J. Colloid Interface Sci., 2005, 286, 90–100.
53. L. Ren, J. Zhang, Y. Li and C. Zhang, Chem. Eng. J., 2011, 168, 553–561.
54. D. Kavitha and C. Namasivayam, Bioresour. Technol., 2007, 98, 14–21.
55. A. Zahangir, S. A. Muyibi and J. Toramae, J. Environ. Sci., 2007, 19, 674–677.
56. M. Sathishkumar, A. R. Binupriya, D. Kavitha, R. Selvakumar, R. Jayabalan, J. G. Choi and S. E. Yun, Chem. Eng. J., 2009, 147, 265–271.
57. K. Yang, W. H. Wu, Q. F. Jing and L. Z. Zhu, Environ. Sci. Technol., 2008, 42, 7931–7936.
58. M. Sathishkumar, A. R. Binupriya, D. Kavitha and S. E. Yun, Bioresour. Technol., 2007, 98, 866–873.
59. J. Pan, X. Zou, X. Wang, W. Guan, Y. Yan and J. Han, Chem. Eng. J., 2010, 162, 910–918.
60. N. Li, Z. Mei and S. G. Ding, J. Inclusion Phenom. Macrocyclic Chem., 2010, 68, 123–129.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7en00787f

‡ Authors Haijian Ou and Weijun Zhang contributed equally to this work.

---