Protonated mesoporous graphitic carbon nitride for rapid and highly efficient removal of microcystins

Abstract

Cyanotoxins have caused worldwide concern due to their diverse occurrence and toxic effects, which has led to an intensive search for cost-effective techniques for their removal from contaminated water. In this study, a novel biomaterial, protonated mesoporous graphitic carbon nitride (mpg-C₃N₄–H⁺) which is fabricated by treating mpg-C₃N₄ with concentrated hydrochloric acid, is applied as a promising bioadsorbent for the uptake of microcystins (MCs). The pH of the reaction media played a significant role in the removal of MCs; maximum adsorption occurred at pH 7. Kinetic studies showed that the adsorption of MC-LR and MC-RR onto the adsorbent was a rapid process. The mpgH⁺ exhibited high adsorption capacities of 2360.96 and 2868.78 μg g⁻¹ from the Langmuir model for MC-LR/RR, respectively. The high adsorption capacity, good solvent stability, and excellent reusability make mpg-C₃N₄–H⁺ promising as a novel adsorbent for the adsorption and removal of MCs from aqueous solution. This information may be useful for further research and practical applications of the novel two-dimensional layered, mesoporous graphitic carbon nitride.

---

1. Introduction

In the past few years, great efforts have been made to study two-dimensional (2D) layered nanomaterials which are one of the most fascinating materials for their high specific surface area and distinct properties.^1,2 As an emerging 2D layered nanomaterial, graphitic-phase carbon nitride (g-C₃N₄) has attracted a great deal of scientific interest because theoretical investigations have revealed outstanding mechanical and thermal properties, and unique electronic and optical properties of g-C₃N₄.^3,4 g-C₃N₄ contains planar layers with π-electron conjugation and is mainly composed of carbon and nitrogen with weak van der Waals forces between C–N layers.⁵ Since g-C₃N₄ is a chemically and thermally stable semiconductor with a band gap of about 2.7 eV, it has been applied in the fields of catalysis,⁶ degradation,⁷ sensing,⁸ drug delivery⁹ and imaging.¹⁰ Very recently, mesoporous graphitic carbon nitride (mpg-C₃N₄) has been prepared by soft chemical routes and methods.¹¹ As a novel kind of g-C₃N₄, mpg-C₃N₄ has a much higher surface area (∼200 m² g⁻¹) than the bulk g-C₃N₄ and extensive further study is valuable.^12,13 However, finer details of the local structure and the composition of mpg-C₃N₄ are lacking in many cases, which also results in limitations and uncertainty in potential applications. To solve these problems, post-functionalization would be a suitable method to improve the properties of mpg-C₃N₄. Among others, direct protonation of base functionalities is a convenient modification route. This is reasonable because direct protonation has been used to disperse and process CNTs.¹⁴ Zhang et al. have reported that carbon nitrides can be reversibly protonated by strong mineral acids, thus modifying the solubility/dispersibility and surface area.¹⁵ Nevertheless, to the best of our knowledge, direct protonation of mpg-C₃N₄ and application of the protonated mpg-C₃N₄ as a bioadsorbent have not been explored until now.

Microcystins (MCs) are known for their rapid activity and acute lethal toxicity, which can cause damage to the liver even at very low concentrations and induce tumour promoting activity through the inhibition of protein phosphatases.¹⁶ To date, over 80 variants of MCs have been discovered. Among them, microcystin-LR (MC-LR) and microcystin-RR (MC-RR) are the most commonly studied MCs. Moreover, a provisional safety guideline of 1.0 μg L⁻¹ MC-LR in drinking water has been recommended by the World Health Organization.¹⁷ Therefore, there is an urgent need to develop a reliable method for rapid and effective removal of MCs from water sources.

To date, various methods have been developed to remove MCs from aqueous solution, such as traditional water treatment technology,¹⁸ chemical oxidation processes,¹⁹ and biological methods.²⁰ However, the above-mentioned methods usually require a high dosage, are time consuming, and may generate toxic disinfection by-products and are thus not viable. More recently, a number of reports have been published on adsorption which is regarded as a simple, effective and time-saving technology for the removal of MCs. Materials including activated carbons (ACs),²¹ clays,²² peats²³ and carbon nanotubes²⁴ have been explored as potential bioadsorbents in the adsorption process and have achieved some success. Pavagadhi et al. have examined the ability of graphene oxide to remove both MC-LR and MC-RR from water.²⁵ Ordered mesoporous silica²⁶ and mesoporous carbons²⁷ were also employed for removal of MCs, with the results showing that mesoporous materials have great potential for uptake of MCs. Xia et al. fabricated metal–organic framework MIL-100 (Al) gels for the adsorption of MC-LR.²⁸ Even so, materials available for highly efficient removal of MCs are still quite limited. Thus, it is still of great significance to discover new materials for highly efficient removal of MCs.

Herein, for the first time, we report a novel bioadsorbent, which was prepared by modification of mpg-C₃N₄ via direct protonation, for rapid and highly efficient removal of MCs (MC-LR and MC-RR, the chemical structures of these two MCs are shown in ESI (Fig. S1†)). We found that the protonation process could not only enhance the adsorption capacities for MCs but also greatly accelerate the removal rate. Moreover, the mpg-C₃N₄–H⁺ shows much higher adsorption capacities than commercial activated carbon in removing MCs. Adsorption kinetics and isotherms were studied, and the experimental results showed that a pseudo-second-order kinetic model and Langmuir isotherms were a better fit for the adsorption of MCs onto the bioadsorbent. The thermodynamic parameters, negative free energy change and negative enthalpy indicate that the adsorption is spontaneous, favorable and exothermic reaction in nature. The result strongly exhibits that mpg-C₃N₄–H⁺ can be successfully applied to convenient, high efficient and fast removal of MCs dissolved in water.

2. Experimental

2.1. Chemicals and reagents

Commercial colloidal silica (Ludox HS40) was purchased from Sigma-Aldrich (St. Louis, Missouri, USA). MC-LR and MC-RR (>95% purity, HPLC) were obtained from Alexis Biochemicals (Switzerland). All reagents were either HPLC grade or analytical grade. Acetonitrile and formic acid were purchased from Fluka. Methanol was obtained from Sigma-Aldrich (HPLC grade). The MC-LR and MC-RR stock solution was prepared at the concentration of 100 mg L⁻¹ in methanol. Deionized water (18.2 MΩ) was prepared using a Milli-Q water purification system (Millipore, USA).

2.2. Apparatus

Scanning electron microscopy (SEM) images were obtained using an FEI Inspect F50 (FEI, USA). Transmission electron microscopy (TEM) analyses were performed on a FEI Tecnai G2 F20 (FEI, USA) at 200 kV. Fourier transform infrared (FT-IR) spectra were obtained using a FT-IR spectrophotometer (Nicolet 6700, Waltham, MA, USA). The surface area and pore diameter were determined using a physisorption analyzer (Micromeritics ASAP 2020 porosimeter, USA) at −196 °C. Before the measurements, samples were degassed in vacuo at 200 °C for at least 480 min. The Brunauer–Emmett–Teller (BET) method was used to calculate the specific surface areas (SBET) using adsorption data at p/p₀ of 0.05–0.3. The pore size distributions (PSDs) were derived from the adsorption branches of isotherms by using the BJH model. The total pore volume (V_t) was estimated from the adsorbed amount at p/p₀ of 0.995. The X-ray powder diffraction (XRD) pattern was determined using a D8 Advance (Bruker, German). Zeta potentials of the adsorbent in ultrapure water were measured on a zeta potential analyzer (Mastersizer 3000, Malvern). Elemental analysis was performed using a varioMICRO cube form Elementar Analysen systeme GmbH.

2.3. Preparation of the mpg-C₃N₄

The mpg-C₃N₄ was prepared according to the previously reported method.¹³ Briefly, 5.0 g of cyanamide was dissolved in a 40% dispersion of 12 nm SiO₂ particles in 12.5 g of water with stirring at 333 K overnight. Then the resulting transparent mixtures were heated for 4 h to 550 °C and kept at this temperature for another 4 h in air. The resulting brown-yellow powder was treated with 4 M ammonium bifluoride for 2 days to remove the silica template. The powders were then filtered and washed with distilled water at least 6 times. Afterwards, the powders were stirred in 0.01 M potassium hydroxide solution for another 2 days. Finally the powders were again filtered and washed at least 3 times with deionized water and 2 times with ethanol. The final product was obtained by drying over night at 100 °C under vacuum.

Protonation of mpg-C₃N₄. 1.0 g of mpg-C₃N₄ was mixed with 10.0 mL of 5.0 M hydrochloric acid. The solution was kept stirring overnight, followed by filtration and washing with deionized water until at neutral pH,and then drying over night at 100 °C under vacuum.

Preparation of bulk g-C₃N₄. The bulk g-C₃N₄ was prepared by polymerization of dicyandiamide molecules at high temperature. In detail, dicyandiamide was heated at 823 K for 4 h in air with a ramp rate of about 2.3 K min⁻¹ for both of the heating and cooling processes. The obtained yellow product was the g-C₃N₄ powder.

2.4. Batch adsorption experiments

Batch adsorption experiments were conducted to assess the adsorption rate and determine the adsorption equilibrium time. To initiate the experiments, 0.5 mg of mpg-C₃N₄–H⁺ was added into a 10.0 mL vial equipped with a Teflon-lined screw cap, then 5.0 mL of 50.0 μg L⁻¹, 100.0 μg L⁻¹, 200.0 μg L⁻¹ MC-LR and MC-RR solution were added separately. Following this the mixture was agitated at 200 rpm and 30 °C in a thermostated rotary shaker (MRC TU400) for 400 min. At each predetermined time point, 1.0 mL of the mixture was taken out. The solutions were separated from the adsorbent by centrifugation at 8000 rpm for 5 min, then the supernatant was filtered through syringe filters (cellulose acetate membranes) with a size of 0.22 μm to remove the particles after the centrifugation, and each residual concentration of MCs was measured. Control experiments were performed with blanks containing no MB under the same conditions as for the MB solution. Each experiment was run in triplicate.

Similarly, adsorption isotherm experiments were conducted to evaluate the maximum adsorption value and the adsorption thermodynamic properties. 0.5 mg of adsorbent was dispersed in 5.0 mL solution with various initial concentrations of MC-LR and MC-RR (20.0–300.0 μg L⁻¹) and agitated for 180 min in a rotary shaker to reach apparent adsorption equilibrium at different temperatures (20–50 °C). Control experiments were preformed with 20.0–300.0 μg L⁻¹ of MCs devoid of adsorbent under the same conditions as with adsorbent. Each experiment was run in triplicate. No noticeable reduction in the initial concentration of MCs studied was observed in the controlled experiments.

The MC solutions were adjusted using either a 1.0 M formic acid or a 1.0 M ammonium hydroxide solution to different pH values (2.0 to 9.0). The effect of ionic strength on MC adsorption to the adsorbent was carried out with background electrolyte (ammonium acetate) concentrations ranging from 0.02 M to 0.1 M. The conditions of the pH dependent adsorption tests and ionic strength dependent adsorption experiments were the same as in the batch adsorption experiments above.

2.5. Analytical methods

The concentration of the MCs was accurately determined using a LC-MS-MS system which consisted of an Accela HPLC system (Thermo Fisher Scientific, USA) with a vacuum degasser, quaternary pump, autosampler and thermostated column compartment, coupled to a TSQ Quantum Access Max™ triple quadrupole mass spectrometer (Thermo Fisher Scientific, USA). The analytical separation of the analytes was achieved using a Hypersil GOLD C18 column (5 μm particle size, 150 × 2.1 mm). The mobile phase was a mixture of acetonitrile (solvent B) and water (45 : 55) containing 0.1% of formic acid (v/v). The flow rate was set at 0.2 mL min⁻¹. The sample injection volume was 10 μL. An isocratic elution was used with a acetonitrile : water ratio of 45 : 55 at the rate of 200 μL min⁻¹. The capillary temperature and vaporizer were set at 350 °C and 300 °C, respectively. Sheath gas pressure and auxiliary gas pressure were carried out at 35 bar and 10 bar separately. The instrument was operated in the positive ion mode. MC-LR and MC-RR were monitored by using the instrument in the SRM mode (m/z 498.5, fragment ion 135.11; m/z 520.0, fragment ion 135.01, respectively).

3. Results and discussion

3.1. Characterization

To investigate the basic composition of the mpg-C₃N₄, XRD was first conducted. The presence of g-C₃N₄ domains is verified by two signals present in the XRD pattern (Fig. 1), namely the strong shoulder peak at 2θ of 27.4° (d = 0.326 nm), which originates from the (002) interlayer diffraction of a CN graphitic-like structure, and the low-angle diffraction peak at 2θ of 13.3° (d = 0.663 nm), which is derived from inplanar repeated tri-s-triazine units. The graphite-like structure of mpg-C₃N₄ is retained after the protonation.

Fig. 1 XRD patterns.

In the FT-IR spectrum (ESI Fig. S2a†), the peaks at 1637 cm⁻¹ and 1243 cm⁻¹ are attributable to the C=N and C–N stretching vibration modes, respectively.²⁹ The peak at 808 cm⁻¹ is related to the s-triazine ring modes.³⁰ After functionalization with strong acids, no peaks corresponding to an amide or –OH group which could be related to broken fragments were found, as previously reported.¹⁵ For example, it could be seen that the peak at 1541 cm⁻¹ disappeared in mpg-C₃N₄–H⁺ (ESI Fig. S2b†). It can also be clearly seen that the main characteristic peaks of the mpg-C₃N₄ appear for the mpg-C₃N₄–H⁺ which further confirms that the core chemical skeleton of mpg-C₃N₄ has remained unchanged after proton treatment using concentrated hydrochloric acid.

Direct evidence of the protonation came from the increased hydrogen content of protonated mpg-C₃N₄, as determined by elemental analysis. A small amount of hydrogen (ca. 2.0%) is most often observed in the starting mpg-C₃N₄ (more appropriately as mpg-C₃N_4.2H_2.0), which could be due to absorbed water and imperfection of the thermolysis. After the protonation, the hydrogen content of protonated mpg-C₃N₄ increased to 2.7% (more appropriately as mpg-C₃N_4.2H_2.7).

The structure and morphologies of the materials were revealed by FESEM and TEM images (Fig. 2). SEM images reveal the typical slate-like, stacked lamellar texture of milled g-C₃N₄, where the lamellar character is indicated by its preferential cleavage planes. Apparently, the protonation turned grey mpg-C₃N₄ (Fig. 2a) into practically white mpg-C₃N₄–H⁺ (Fig. 2b) and the surface quality of the composites is improved evidently. TEM images indicate that a lot of mesoporous exist in the material which can be attributed to the voids obtained after etching with ammonium bifluoride (Fig. 2c and d). The zeta-potential of mpg-C₃N₄–H⁺ dispersions in water is shifted after contact with concentrated hydrochloric acid from negative to positive surface charges at almost every pH value, as shown in Fig. S4,† which also demonstrates successful protonation.

Fig. 2 (a) SEM image for mpg-C₃N₄; (b) SEM image for mpg-C₃N₄–H⁺; (c) TEM image for mpg-C₃N₄; (d) TEM image for mpg-C₃N₄–H⁺.

The N₂ adsorption isotherms measured for g-C₃N₄, mpg-C₃N₄ and mpg-C₃N₄–H⁺ resemble type IV with H3-type hysteresis loops (Fig. 3), which confirms the presence of interconnected mesopores. The mpg-C₃N₄ shows a much higher BET surface area than g-C₃N₄. Correspondingly, similar distribution curves were observed for the pore-size of mpg-C₃N₄ and mpg-C₃N₄–H⁺. In addition, after the protonation, mpg-C₃N₄–H⁺ exhibits a slightly larger total adsorption average pore width, total pore volume, and the experimental multipoint BET surface area of 14.68 nm, 0.71 cm³, 193.52 m² g⁻¹ than mpg-C₃N₄ of 13.65 nm, 0.65 cm³, 189.78 m² g⁻¹, respectively.

Fig. 3 (a) N₂ adsorption–desorption isotherms; (b) pore diameter distribution profiles.

3.2. Effect of different factors on adsorption

Effect of initial solution pH. The experiments were carried out in the pH range 2.0–9.0 to examine the effect of pH on the adsorption of MC-LR/RR by mpg-C₃N₄–H⁺, and the results are illustrated in Fig. 4a. The results show that the adsorption of MC-LR and MC-RR followed the same trend: the adsorption capacity of mpg-C₃N₄–H⁺ for MCs increased as the pH increased from 2.0 to 7.0, then decreased with further increase of pH from 7.0 to 9.0. Therefore, it is clear that the maximum adsorption is observed at pH 7.0. Therefore, all other experiments were carried out at pH 7.0.

Fig. 4 (a) Effect of pH on the adsorption of MCs (25.0 μg L⁻¹) on mpg-C₃N₄–H⁺ (0.1 mg); (b) effect of the concentration of ammonium acetate on the adsorption of MCs (50.0 μg L⁻¹) on mpg-C₃N₄–H⁺ at 30 °C and pH 7.0.

Since the structure of MCs contains numerous ionizable groups, the overall charge on the toxin is pH dependent. Maagd et al. demonstrated that the MC-LR species remain neutral at the narrow pH range of 2.09–2.19; MC-LR is protonated to cationic species [(COOH)₂(NH₂⁺)] when the pH < 2.09 and deprotonated to anionic species, [(COO⁻)₂(NH)] and [(COO⁻)₂(NH₂⁺)] at pH > 2.19.³¹ Therefore, since MC-RR and MC-LR share similar structures, it can be inferred that they would behave in a similar way. As the solution pH increased from 2.0 to 9.0, the positive charges on the surface of mpg-C₃N₄–H⁺ decreased, i.e. less positive zeta potential (ESI Fig. S4†), which would have a reduced effect on the adsorption of negatively charged MCs due to electrostatic attraction. Whereas, as the solution pH increased from 2.0 to 9.0, more and more MCs were deprotonated to form COO⁻ groups with a negative charge, thereby enhancing the electrostatic attraction between negatively charged MC anions and the positively charged surface of the mpg-C₃N₄–H⁺. The above two opposing factors led to the optimal pH at 7.0.

Furthermore, the partitioning effect of charged MC-LR and MC-RR was employed in the present study to explain the decreased adsorption of MC-LR at pH > 7.0. Partitioning is the measure of the differential solubility of a compound between two immiscible solvents at equilibrium. Normally, one of the solvents is water and the second one is a hydrophobic or nonpolar solvent such as octanol. It correlates with the tendency of a molecule to concentrate in the lipids of organisms and the organic carbon of sediments and soils. As the pH increases from 1 to 10, the n-octanol/water distribution ratio D_ow for MC-LR decreases from 2.18 to 1.76, revealing that MC-LR exhibits increased hydrophilicity and consequently decreased partitioning in octanol with the increase in pH. Therefore, with the increase in pH, MC-LR has a tendency to remain in the aqueous phase rather than undergoing adsorption. It could be inferred that MC-RR would show a very similar kind of response as MC-LR in terms of partitioning due to the similarities between their structures.

Effect of ionic strength. Generally, there are various salts and metal ions in natural water. In order to investigate the effect of coexisting ions on adsorption properties of mpg-C₃N₄–H⁺, different concentrations of ammonium acetate were added into the adsorption system. As seen in Fig. 4b, MC uptake capacities decreased significantly once 0.01 mol L⁻¹ ammonium acetate is added, furthermore, the MC removal efficiency dropped persistently with more and more electrolyte in the solution. The above results indicate that electrostatic interaction is one of the mechanisms for the adsorption of MCs on the adsorbent.

As the results in Fig. 4 show, the mpg-C₃N₄–H⁺ has a better adsorption capacity of MC-RR than MC-LR, which is in agreement with that reported previously.³² It may be attributed to their different molecular compositions. It has been generally accepted that the difference in adsorption capacity mainly comes from the arginine unit instead of leucine at the second position in MC-RR. Hence, there is an increased tendency to form cationic bridging and more hydrogen bonds between the adsorbents and MC-RR.³²

Comparison of the adsorption capacity. In order to further study the mechanisms of adsorption of MCs on mpg-C₃N₄–H⁺, the adsorbed quantity of MCs by three adsorbents is compared in Fig. 5. It shows that mpg-C₃N₄–H⁺ has a better MC removal efficiency than mpg-C₃N₄ at different times and the adsorption capacity of mpg-C₃N₄ is much better than that of g-C₃N₄. The uptake efficiency of mpg-C₃N₄–H⁺ can be as high as 96% for both MC-RR and MC-LR after contact for 15 min at the tested concentration. It is clear that the protonation not only enhances the ability of adsorption for MCs but also greatly accelerates the removal rate of MCs.

Fig. 5 Adsorption isotherms of MCs (50.0 μg L⁻¹) over the three adsorbents at 30 °C and pH 7.0.

The BET surface area and pore volume of mpg-C₃N₄ are 189.78 m² g⁻¹ and 0.65 cm³ g⁻¹, respectively, far beyond those of non-mesoporous g-C₃N₄, which are 13.57 m² g⁻¹ and 0.27 cm³ g⁻¹, respectively. The adsorption amount increases by a factor of 2 for MC-LR and 1.2 for MC-RR with the increase in surface area and pore volume. Therefore, the adsorbents with larger surface area and pore volume manifest higher adsorption capacities of MC molecules than those with low ones.

Furthermore, we can obtain from Fig. S4† that the zeta potential of the materials (0.15 mg mL⁻¹) increases from −22.21 mV to +17.07 mV after protonation at pH 7.0. Correspondingly, the adsorption equilibrium time decreases from about 100 min to 30 min (Fig. 5), the adsorption amount of MCs leveled off at about 485.0 μg g⁻¹, revealing that the protonation process greatly enhances the uptake efficiency of MCs onto the adsorbent. It also demonstrates that electrostatic interaction plays an important role in adsorption of MCs on the adsorbent.

3.3. Kinetics for the adsorption

The time-dependent adsorption capacity was obtained to study the kinetics for the adsorption of MCs on mpg-C₃N₄–H⁺ (Fig. 6a and b). It is observed that the removal of MCs from aqueous solution is associated with an extremely rapid initial stage and the adsorption equilibrium is achieved within 15 min for both MC-LR/RR at 50.0 μg L⁻¹. The equilibrium adsorption capacity increased from 474.9 to 1483.6 μg g⁻¹ for MC-LR and 481.4 to 1949.0 μg g⁻¹ for MC-RR as the initial concentration varied from 50.0 to 200.0 μg L⁻¹. Furthermore, the adsorption capacity significantly increased as the initial concentration of MCs increased, indicating the favorable adsorption at high concentrations of MCs.

Fig. 6 (a) and (b) Effect of contact time on the adsorption of MCs on mpg-C₃N₄–H⁺ (0.1 mg) at different initial concentrations of MCs at 30 °C and pH 7.0; (c) and (d) plots of pseudo-second-order kinetics for the adsorption of MCs on mpg-C₃N₄–H⁺ (0.1 mg).

Three of the most widely used kinetic models, i.e. the pseudo-second-order equation, pseudo-first-order equation and intra-particle diffusion model, were used to examine the adsorption kinetic behavior of MCs onto mpg-C₃N₄–H⁺. The best-fit model was selected based on the linear regression correlation coefficient values (R²). The result is shown in Table 1.

Table 1 Kinetic parameters for the adsorption of MCs on mpg-C₃N₄–H⁺ at 30 °C

	C0/(ppb)	qe(exp)/(μg g^−1)	Pseudo-second-order kinetic model			Intra-particle diffusion model
			qe(cal)/(μg g^−1)	k2/(μg g^−1 min^−1)	R^2	Kid/(μg g^−1 min^−1/2)	C (μg g^−1)	R^2
MC-LR	50.0	474.87	476.19	1.04 × 10^−3	1.0000	15.41	372.52	0.5796
	100.0	803.59	806.45	3.82 × 10^−4	1.0000	63.23	427.36	0.5664
	200.0	1483.61	1484.42	2.99 × 10^−4	1.0000	144.19	663.01	0.9986
MC-RR	50.0	481.44	480.77	3.01 × 10^−2	1.0000	0.06	479.56	0.4393
	100.0	967.69	970.87	4.74 × 10^−4	1.0000	126.97	239.53	0.9842
	200.0	1949.04	1956.97	1.13 × 10^−4	1.0000	189.22	501.67	0.9885

---

The pseudo-first-order kinetic model might be represented by eqn (1).

ln(q_e − q_t) = ln q_e − k₁t (1)

where k₁: the apparent pseudo-first-order constant (min⁻¹); q_e: amount adsorbed at equilibrium (μg g⁻¹); q_t: amount adsorbed at time t (μg g⁻¹); t: adsorption time (min). Therefore, the first order kinetic constant (k₁) can be calculated by k₁ = −slope when ln (q_e − q_t) is plotted against t.

The adsorption data were also analyzed using the versatile pseudo-second-order kinetic model:

where k₂: the apparent pseudo-second-order rate constant (g μg⁻¹ min⁻¹) Similarly, the second-order kinetic constant (k₂) can be calculated by the values of 1/q_e and 1/k₂q_e² when t/q_t is plotted against t.

An intra-particle mass transfer diffusion model proposed by Weber and Morris can be written as follows:

q_t = K_idt^1/2 + C (3)

where q_t (mg g⁻¹) is the adsorption capacity at time t, K_id (μg g⁻¹ min^−1/2) is the intraparticle diffusion rate constant, and C is the intercept. The K_id value can be found from the slope of the q_t against t^1/2 plot.

The pseudo-first-order kinetic model data is shown in ESI Table 1.† The correlation coefficient values are very low suggesting that pseudo-first-order kinetic model was not suitable to be chosen to describe the adsorption. All the experimental data showed better compliance with a pseudo-second-order kinetic model in terms of higher correlation coefficient values (R² > 0.9999) and closer values between q_e,cal and q_e,exp (Fig. 6c and d). It is clear that the adsorption of MCs on the adsorbent is well described by a versatile pseudo-second-order kinetic model which is based on the adsorption capacity on the solid phase. Moreover, the rate constant k₂ decreased as the initial MC concentration increased, indicating that the rate-limiting step might be chemical adsorption, involving valency forces through sharing or exchange of electrons between anions and the adsorbent, while chemisorption is a kind of adsorption which involves a chemical reaction between the adsorbent surface and the adsorbate.³³ Besides, the possibility of establishing π–π interactions between the benzene rings in MCs and mpg-C₃N₄–H⁺ also plays an important role.

For the intra-particle mass transfer diffusion model, if the value of C is zero, then the rate of adsorption is controlled solely by intra-particle diffusion for the entire adsorption period.³⁴ However all the linear portions don’t pass through the origin (ESI Fig. S6†) and the plot of q_t versus t^1/2 shows multilinear portions, suggesting that more than one process affects the adsorption. All of the intercepts of the plots reflected an obvious boundary layer effect (Table 1), revealing that film (boundary layer) diffusion controlled the adsorption rate at the beginning,³⁵ but intra-particle diffusion is not solely the rate-controlling step, for the larger the intercept, the greater the contribution of the surface adsorption in the rate-controlling step.

3.4. Isotherm modelling

The adsorption isotherms of MCs on mpg-C₃N₄–H⁺ were measured at four different temperatures (ESI Fig. S7†). In order to describe the adsorption isotherm more scientifically, the Langmuir and Freundlich models were selected for this study. The linearized Langmuir isotherm eqn (4) and Freundlich isotherm eqn (5) can be expressed as follows:

q_e = K_FC^1/n_e (5)

where C_e: equilibrium concentration of adsorbate (μg L⁻¹); q_e: amount adsorbed at equilibrium (μg g⁻¹); q_m: maximum adsorption capacity (μg g⁻¹); b: Langmuir constant (L μg⁻¹ or L mol⁻¹); K_F: the Freundlich adsorption constant (μg g⁻¹(L μg⁻¹)^1/n); 1/n: another constant related to the surface heterogeneity (unitless). The slope and intercept of linear plots of C_e/q_e against C_e yield the values of 1/q_m and 1/q_mb for eqn (4) and the slope and intercept of linear plots of ln q_e against ln C_e yield the values of 1/n and ln K_F for eqn (5).

The theoretical parameters of adsorption isotherms along with regression coefficients (R²) are summarized in Table 2. R² values for the Langmuir model are higher than for the Freundlich model showing that the Langmuir model is a reasonably better fit with the adsorption process. It is clear that the correlation coefficients for MCs are high, indicating that the MCs adsorbed on the surface of adsorbent is a monolayer coverage. It is notable that an increase in temperature resulted in a corresponding decrease in adsorption capacity of MCs, which showed that uptake of MCs onto the mpg-C₃N₄–H⁺ is an exothermic process. The mpg-C₃N₄–H⁺ showed a very high adsorption capacity of 2360.96 and 2868.78 μg g⁻¹ from the Langmuir model for MC-LR and MC-RR, respectively, while the maximum adsorption capacity obtained with the commercial activated carbon was 1481.7 μg g⁻¹ and 1034.1 μg g⁻¹ for MC-LR and MC-RR, respectively.³⁶

Table 2 Isotherm model constants and regression coefficients for MCs adsorption onto mpg-C₃N₄–H⁺

	T (K)	Langmuir isotherm constants			Freundlich isotherm constants
		qm/(μg g^−1)	KL (L μg^−1)	R^2	1/n	KF (μg g^−1(L μg^−1)^1/n)	R^2
MC-LR	293.0	2360.96	0.05	0.9980	0.59	140.89	0.9697
	303.0	2320.19	0.05	0.9986	0.62	104.56	0.9663
	313.0	2315.47	0.04	0.9962	0.56	185.94	0.9493
	323.0	2249.75	0.03	0.9988	0.57	163.89	0.9529
MC-RR	293.0	2868.78	0.08	0.9988	0.60	253.64	0.9453
	303.0	2685.65	0.07	0.9978	0.58	231.59	0.9493
	313.0	2534.80	0.06	0.9968	0.57	205.55	0.9589
	323.0	2360.56	0.05	0.9946	0.56	186.29	0.9586

---

3.5. Thermodynamics for the adsorption

In the process of MC adsorption, according to the adsorption equilibrium at different temperatures, the thermodynamic parameters, standard free energy change (ΔG°, kJ mol⁻¹), entropy change (ΔS°, J mol⁻¹ K) and enthalpy change (ΔH°, kJ mol⁻¹) can be estimated from the following relationship (ESI Fig. S8†):

ΔG = RT ln b (6)

(where R is the gas constant).

The Langmuir constant b (dimension: L mol⁻¹) can be obtained from the slope/intercept of the Langmuir plot. The enthalpy change ΔH values for MC adsorption over mpg-C₃N₄–H⁺ were both negative, −21.28 kJ mol⁻¹ for MC-LR and −12.56 kJ mol⁻¹ for MC-RR, confirming endothermic adsorption in accordance with the decreasing adsorption capacity associated with increasing adsorption temperature. This result may be attributed to the high temperature which will block the interaction between MCs and the adsorbent. The entropy changes ΔS were −29.58 J mol⁻¹ K⁻¹ and 3.78 J mol⁻¹ K⁻¹ for MC-LR and MC-RR, respectively. The positive ΔS means an increased randomness with the adsorption of MC-RR probably because the number of desorbed water molecules is larger than that of the adsorbed MC-RR molecules. On the contrary, the negative ΔS for MC-LR means a decreased randomness with adsorption. According to eqn (7), the free energies of adsorption at 20 °C, 30 °C, 40 °C and 50 °C were −12.49, −12.42, −12.24, −11.53 kJ mol⁻¹ for MC-LR and −13.64, −13.77, −13.73, −13.77 kJ mol⁻¹ for MC-RR, respectively (Table 3). These negative free energies confirmed that adsorption is spontaneous under the experimental conditions used.

Table 3 Thermodynamic parameters for adsorption

	T/(K)	Thermodynamic parameters
		ΔG°/(kJ mol^−1)	ΔH°/(kJ mol^−1)	ΔS°/(J mol^−1 K^−1)
MC-LR	293.0	−12.49	−21.28	−29.58
	303.0	−12.42
	313.0	−12.24
	323.0	−11.53
MC-RR	293.0	−13.64	−12.56	3.78
	303.0	−13.77
	313.0	−13.73
	323.0	−13.77

---

3.6. Desorption and regeneration

To evaluate the possibility of regeneration and reusability of mpg-C₃N₄–H⁺ as an adsorbent, desorption experiments were conducted. Desorption and regeneration experiments were achieved by using ethanol under ultrasound assisted desorption for 15 min. The effect of two consecutive adsorption–desorption cycles was studied (Fig. 7). The MC removal efficiency almost keeps a steady value and it is still above 90% after recycling 2 times. These results demonstrate that the mpg-C₃N₄–H⁺ can be regenerated and is suitable for adsorptive MC removal.

Fig. 7 (a) Effect of contact time on the MC-LR (200.0 μg L⁻¹) adsorption at 30 °C and pH 7.0; (b) effect of contact time on the MC-RR (200.0 μg L⁻¹) adsorption at 30 °C and pH 7.0; (c) and (d) pseudo-second-order plots to show the re-usability of the mpg-C₃N₄–H⁺ in adsorption of MC-LR/RR.

A FT-IR spectrum and an SEM image of the recycled mpg-C₃N₄–H⁺ are shown in Fig. S2 and S5,† respectively. From Fig. S2,† it can be seen that the characteristic peaks of the recycled mpg-C₃N₄–H⁺ are very close to the original mpg-C₃N₄–H⁺, which further confirms that the core chemical skeleton of mpg-C₃N₄–H⁺ has remained unchanged after exposure to MCs. Based on Fig. S5,† it can also be observed that the typical slate-like structure still remains in mpg-C₃N₄–H⁺. All of these results can help to draw the conclusion that the structure of the mpg-C₃N₄–H⁺ is stable during the adsorption process.

4. Conclusion

In this work, we reported the adsorption of MCs on a highly protonated mesoporous graphitic carbon nitride and studied the adsorption kinetics, isotherms, thermodynamics and regeneration of the sorbent. The measurements indicated that the mpg-C₃N₄ could be protonated by a convenient modification route. It is found that the removal efficiency of MCs through mpg-C₃N₄–H⁺ is much better than mpg-C₃N₄ and g-C₃N₄. The results also confirm that mpg-C₃N₄–H⁺ can be used in place of activated carbon since it has a higher adsorption capacity and is more efficient at removing MCs. The good results demonstrated the potential of mpg-C₃N₄–H⁺ in adsorption and removal of MCs from aqueous solution.

Acknowledgements

The authors are grateful for the National Nature Sciences Foundation of China (21275029 and 201405018), the National Basic Research Program of China (No. 2010CB732403), the “12th Five-Year National Science and Technology Support Program” (2012BAD29B06), and the Program for Changjiang Scholars and Innovative Research Team in University (no. IRT1116).

References

1. K. P. Loh, Q. Bao, G. Eda and M. Chhowalla, Nat. Chem., 2010, 2, 1015–1024.
2. L. Feng, L. Wu and X. Qu, Adv. Mater., 2013, 25, 168–186.
3. S. Yang, Y. Gong, J. Zhang, L. Zhan, L. Ma, Z. Fang, R. Vajtai, X. Wang and P. M. Ajayan, Adv. Mater., 2013, 25, 2452–2456.
4. Y. Zhang and M. Antonietti, Chem.–Asian J., 2010, 5, 1307.
5. M. Groenewolt and M. Antonietti, Adv. Mater., 2005, 17, 1789–1792.
6. X. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M. Carlsson, K. Domen and M. Antonietti, Nat. Mater., 2008, 8, 76–80.
7. X. Wang, X. Chen, A. Thomas, X. Fu and M. Antonietti, Adv. Mater., 2009, 21, 1609–1612.
8. C. Cheng, Y. Huang, J. Wang, B. Zheng, H. Yuan and D. Xiao, Anal. Chem., 2013, 85, 2601–2605.
9. L.-S. Lin, Z.-X. Cong, J. Li, K.-M. Ke, S.-S. Guo, H.-H. Yang and G.-N. Chen, J. Mater. Chem. B, 2014, 2, 1031–1037.
10. X. Zhang, H. Wang, H. Wang, Q. Zhang, J. Xie, Y. Tian, J. Wang and Y. Xie, Adv. Mater., 2014, 26, 4438–4443.
11. X. Wang, K. Maeda, X. Chen, K. Takanabe, K. Domen, Y. Hou, X. Fu and M. Antonietti, J. Am. Chem. Soc., 2009, 131, 1680–1681.
12. K. Takanabe, K. Kamata, X. Wang, M. Antonietti, J. Kubota and K. Domen, Phys. Chem. Chem. Phys., 2010, 12, 13020–13025.
13. F. Su, M. Antonietti and X. Wang, Catal. Sci. Technol., 2012, 2, 1005–1009.
14. S. Ramesh, L. M. Ericson, V. A. Davis, R. K. Saini, C. Kittrell, M. Pasquali, W. Billups, W. W. Adams, R. H. Hauge and R. E. Smalley, J. Phys. Chem. B, 2004, 108, 8794–8798.
15. Y. Zhang, A. Thomas, M. Antonietti and X. Wang, J. Am. Chem. Soc., 2008, 131, 50–51.
16. P. Shen, Q. Shi, Z. Hua, F. Kong, Z. Wang, S. Zhuang and D. Chen, Environ. Int., 2003, 29, 641–647.
17. I. R. Falconer, Environ. Toxicol., 1999, 14, 5–12.
18. T. Jurczak, M. Tarczynska, K. Izydorczyk, J. Mankiewicz, M. Zalewski and J. Meriluoto, Water Res., 2005, 39, 2394–2406.
19. S. Merel, B. LeBot, M. Clément, R. Seux and O. Thomas, Chemosphere, 2009, 74, 832–839.
20. D. G. Bourne, R. L. Blakeley, P. Riddles and G. J. Jones, Water Res., 2006, 40, 1294–1302.
21. C. Svrcek and D. W. Smith, J. Environ. Eng. Sci., 2004, 3, 155–185.
22. R. J. Morris, D. E. Williams, H. A. Luu, C. F. Holmes, R. J. Andersen and S. E. Calvert, Toxicon, 2000, 38, 303–308.
23. M. Sathishkumar, S. Pavagadhi, K. Vijayaraghavan, R. Balasubramanian and S. Ong, J. Hazard. Mater., 2010, 184, 417–424.
24. H. Yan, A. Gong, H. He, J. Zhou, Y. Wei and L. Lv, Chemosphere, 2006, 62, 142–148.
25. S. Pavagadhi, A. L. L. Tang, M. Sathishkumar, K. P. Loh and R. Balasubramanian, Water Res., 2013, 47, 4621–4629.
26. W. Teng, Z. Wu, D. Feng, J. Fan, J. Wang, H. Wei, M. Song and D. Zhao, Environ. Sci. Technol., 2013, 47, 8633–8641.
27. W. Teng, Z. Wu, J. Fan, H. Chen, D. Feng, Y. Lv, J. Wang, A. M. Asiri and D. Zhao, Energy Environ. Sci., 2013, 6, 2765–2776.
28. W. Xia, X. Zhang, L. Xu, Y. Wang, J. Lin and R. Zou, RSC Adv., 2013, 3, 11007–11013.
29. Y. Zhao, D. Yu, H. Zhou, Y. Tian and O. Yanagisawa, J. Mater. Sci., 2005, 40, 2645–2647.
30. X. Li, J. Zhang, L. Shen, Y. Ma, W. Lei, Q. Cui and G. Zou, Appl. Phys. A: Mater. Sci. Process., 2009, 94, 387–392.
31. P. De Maagd, A. J. Hendriks, W. Seinen and D. T. Sijm, Water Res., 1999, 33, 677–680.
32. X. Wu, B. Xiao, R. Li, C. Wang, J. Huang and Z. Wang, Environ. Sci. Technol., 2011, 45, 2641–2647.
33. G. Crini, H. N. Peindy, F. Gimbert and C. Robert, Sep. Purif. Technol., 2007, 53, 97–110.
34. V. Vadivelan and K. V. Kumar, J. Colloid Interface Sci., 2005, 286, 90–100.
35. L. Wang, J. Zhang, R. Zhao, C. Li, Y. Li and C. Zhang, Desalination, 2010, 254, 68–74.
36. S. Pavagadhi, A. L. Tang, M. Sathishkumar, K. P. Loh and R. Balasubramanian, Water Res., 2013, 47, 4621–4629.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details and additional figures. See DOI: 10.1039/c5ra01415h

---