Efficient removal of three dyes using porous covalent triazine frameworks: adsorption mechanism and role of pore distribution

Dyes are widely used in various fields, including textile, leather, paper, plating, medicine and food industries etc. (Guo et al. 2014; Yu et al. 2015; Niu et al. 2018; Zhang et al. 2018). A large amount of wastewater was produced in the process of production and use, which caused serious environmental pollution and ecological harm (Singh et al. 2015; Narvekar et al. 2018). In addition, dyes affect the photosynthetic activities of aquatic species due to the color stability, and some of them have high toxicity, carcinogenic, teratogenic and mutagenic effects, and alter gene expression by epigenetic mechanisms (Yao et al. 2010; Perullini et al. 2014; Ojedokun & Bello 2017; Muthusaravanan et al. 2018; Sharavanan et al. 2019). Therefore, removal of dyes from wastewater has attracted much attention for decades. Various methods such as adsorption, flocculation, electrochemical, photocatalysis, advanced oxidation processes and biological methods have been used for treatment of dyes in water solution (Park et al. 2012; Sattarahmady et al. 2013; Konicki et al. 2017; Moreira et al. 2017; Chen et al. 2018; Gu et al. 2018; Katheresan et al. 2018; Zhao et al. 2018).

Adsorption treatment has been recognized as a simple and effective method to remove dyes from aqueous solution, due to its high efficiency, convenience, low cost and easy desorption (Ahmed et al. 2020; Navya et al. 2020). Many studies have reported the adsorption of dyes on carbon-based materials (Al-Degs et al. 2009; Asouhidou et al. 2009; Ip et al. 2010; Gupta & Khatri 2019; Streit et al. 2019; Figueiredo et al. 2011). Liu et al. reported that ordered mesoporous carbon displayed much stronger adsorption than activated carbon (Liu et al. 2009). Gupta & Khatri reported that the specific area and surface polarity conducted fast and efficient adsorptive removal of organic dyes (Gupta & Khatri 2019). Therefore, carbon-based materials with enlarged pore size avoid the size exclusion effect, leading to relatively high adsorption capacity especially for bulky pollutants.

Covalent triazine-based frameworks (CTFs) were first reported in 2008 (Kuhn et al. 2008a). They have attracted great attention for energy gas storage and catalytic support materials (Chan-Thaw et al. 2010; Li et al. 2018; Jena et al. 2018), because of the very large specific surface area, homogeneous pore structure, and high chemical and thermal stability (Kuhn et al. 2008a, 2008b; Kuhn et al. 2009). We reported that a synthesized CTF showed high adsorption affinity, fast adsorption/desorption kinetics, and complete adsorption reversibility for monoaromatic compounds (Liu et al. 2012). In addition, CTFs displayed high adsorption affinity for organic pollutants in previous studies (Liu et al. 2012, 2013, 2015, 2017; Zhang et al. 2011, 2020). These properties make CTFs a promising adsorbent for the removal of organic contaminants in water treatment. However, to the best of our knowledge, little study has reported the adsorption of dyes on CTFs and the adsorption mechanisms and the role of pore distribution were not well understood.

The objective of this work is to explain the adsorption mechanism for three dyes on CTFs and study the roles of CTF pore structure for adsorption. The synthesized CTFs were characterized by XRD, FTIR and N₂ adsorption/desorption isotherms. Pore size distribution and the effect of pH were further performed to understand the mechanisms.

Methylene blue (MB) and methyl orange (MO) were purchased from Tianjin Institute of Chemical Co., Ltd. Chrysophenine G (CG) was supplied by Tokyo Chemical Industry. 1,4-Dicyanobenzene and 4.4′-biphyldicarbonitrile were purchased from purchased from Aldrich. Chemical structures of MB, MO and CG are presented in Supporting Information (SI) Figure S1.

Two different CTFs (named CTF-1 and CTF_DCBP) were synthesized using monomer 1,4-dicyanobenzene and 4.4′-biphenyldicarbonitrile according to literature method (Kuhn et al. 2008a). The details of the synthesis are listed in SI Text S1 and the structures of CTF-1 and CTF_DCBP are presented in SI Figure S2.

X-ray diffraction (XRD) pattern of CTFs was collected from a RigakuD/max-RA powder diffraction-meter using Cu Kα radiation (Rigaku, Japan). Elemental analysis of CTFs was performed in an elemental analyzer of Vario MICRO (Elementar, Germany). Transmission electron microscope (TEM) images of the CTFs were obtained from a JEM-2100 transmission electron microscope. Transmission Fourier transform infrared (FTIR) spectrums of the CTFs were measured using a Nexus 870 spectrometer (Nicolet, USA USA). BET and pore size distribution were obtained on a Micromeritics ASAP 2020 apparatus at −196 °C (Micromeritics Instrument Co., USA).

Adsorption isotherms were conducted by batch adsorption experiment. Briefly, adsorptions (20 mg of CTF-1 and 10 mg of CTF_DCBP) were introduced into 40 ml glass vials equipped with polytetrafluoroethylene-lined receiving 40 mL of dyes solution with varied intial concentrations (background solution: 0.02 M NaCl). The samples were mixed using an orbital shaker protected from light at room temperature for 48 h. After filtration, the solute was analyzed using UV–vis spectrometer with detecting wavelengths at 664, 462 and 400 nm for MB, MO and CG, respectively. All adsorption experiments were conducted in duplicate.

Separate sets of experiments were performed to test the effects of pH and ionic strength. In the pH experiments, 10 mg CTF-1 and 40 ml of 50 mg/L MB solutin (80 mg/L MO and 12.5 mg/L CG) with different pH was pread-justed using 0.1 M HCl and NaOH to ensure the desired pH (3–11) at adsorption equilibrium. In the ionic strength experiments, adsorption was performed using background solutions (0.01, 0.02, 0.05, and 0.1 M) NaCl or CaCl₂ with 10 mg CTF-1 and 40 ml of 90 mg/L MB solutin (75 mg/L MO and 20 mg/L CG).

The pore size distribution of the adsorbents before and after adsorption were collected by N₂ adsorption/desorption isotherms. The adsorbents after adsorption filtered with 0.45 μm fiber filter and dried at 40 °C using a vacuum drying oven. Prior to determine pore size distribution, the adsorbents (CTF-1 and CTF_DCBP) were activated at 200 °C, and the adsorbents after loading the dyes were activated at 80 °C.

XRD patterns of the CTFs are displyed in SI Figure S3. The diffraction peaks at 2θ 7.2° and 15.1° for CTF-1, reflecting the crystalline triazine-based organic framework with hexagonal pores (Kuhn et al. 2008a). Compared with XRD patterns of the CTF-1, CTF_DCBP showed high disordered crystal because of the salt templating effect (Kuhn et al. 2009). Nitrogen adsorption/desorption isotherms of CTFs are presented in SI Figure S4. Obviously, typical capillary condensation was observed within a relative pressure range of 0.2–0.9 for adsorbent of CTF_DCBP, reflecting the presence of mesoporous. TEM images of CTFs are displayed in SI Figure S5. Similar morphologies are obtained for the CTF-1 and CTF_DCBP. CTFs presents a homogeneous structure with microporosity, while the structure of CTF_DCBP becomes rough with increasing mesopore size. BET and pore volume of CTFs are listed in Table 1. The BET surface area of the CTF-1 and CTF_DCBP were 782.44 and 1,745.45 m²/g, and the pore volume were 0.4 and 1.42 cm³·g⁻¹. In addition, micropores were the major pores of CTF-1, while both micropores and mesopores were detected in CTF_DCBP. The transmission FTIR spectrum of CTFs is shown in Figure 1. As shown in Figure 1, the strong peaks around 1,352 and 1,507 cm⁻¹ are assigned to vibrations of the triazine structure (Bojdys et al. 2010).

where q_e (mmol/g) and C_e (mmol/L) are the adsorbed concentration and the aqueous concentration at adsorption equilibrium, q_m(mmol/g) is the adsorption capacity, and b (L/mmol) is the Langmuir affinity coefficient, K_F (mg^1–n Lⁿ/g) is the Freundlich affinity coefficient; n is the Freundlich linearity index. The fitting parameters are summarized in SI Table S2. Adsorption of three dyes to the CTFs can be well described by the Freundlich model with R² higher than 0.93. In addtion, the linearity indexes (n) are smaller than 1, suggesting adsorption heterogeneity.

The cavity of CTF is composed of benzene rings and triazine rings that form a large conjugated chain structure, which is rich in π electrons and can be used as π-electron donor (Liu et al. 2012). MB as a cationic dye has protonated N atoms. Therefore, cation-π interaction mechanism is generated between protonated N in MB and π-electrons in the surface of CTFs, leading the strong adsorption affinity of MB on CTFs. It is note that adsorption of MO on CTF-1 and CTF_DCBP was close to MB, but adsorption mechanism is different. Electrostatic interaction is likely responsible for the strong adsorption of the dye MO. As shown in SI Figure S6, the point of zero charge (PZC) for CTF-1 and CTF_DCBP are at pH 7.2 and 8.0. Therefore, the CTFs is positively charged and forming electrostatic interaction with anionic MO. In addtion, the triazine structure has a strong ability to form hydrogen bonds, and hydrogen bonding energy between 1,3,5-triazine and water calculation varied from 3.38 to 22.60 kJ·mol⁻¹ as the report (Li et al. 2007). Hydrogen bonding interaction between polar functional group and triazine structure further enhanced the adsorption of MO CTFs. Similar results reported that hydrogen bonds formed between MO and the surface of biopolymer chitin and water/1,2-dichloroethane (Longhinotti et al. 1998; Rinuy et al. 2000).

Unit surface area-based adsorption data are compared in Figure 3 to further explain the adsorption patterns between CTF-1 and CTF_DCBP. As shown in Figure 3, adsorption for MB and MO on CTF-1 is higher than on CTF_DCBP, particularly at low concentration. The strong adsorption for MB and MO on CTF-1 may be caused by the micropore-filling mechanism. Many previous studies (Nguyen et al. 2007; Ji et al. 2009; Liu et al. 2012; Liu et al. 2013) have reported this mechansiam. Such as Nguyen et al. (Nguyen et al. 2007) reported that micropore-filling mechanism enhanced aromatic hydrocarbons adsorption on carbonaceous adsorbents. This agrees well with the pore volume results (see SI Table S1). However, adsorption for CG on CTF-1 is dramatically suppressed by the size exclusion effect due to its large molecules. CTT_DCBP displayed higher normalized adsorption than CTF-1 for CG in Figure 3. This attributed to the larger mesopore volume of CTF_DCBP, which effectively avoids the size exclusion effect. The importance of pore structure is further discussed in pore size distribution (see below).

Pore size distribution of CTFs before and after dyes adsorption are compared in Figure 4 and Table 1 to illustrate the impact of adsorbent pore size and adsorbate molecular size. For CTF-1 in Table 1, the total pore volume of CTF-1 decreased from 0.40 cm³/g to 0.160, 0.179 and 0.316 cm³/g after adsorption of MB, MO and CG. The slightly change of total pore volume for CG is attribute to large size CG, leading to only adsorption on the external surface of CTF-1 because of the size exclusion effect. However, for CTF_DCBP in Figure 4(b), after adsorption of CG most pore diameters around 2 to 4 nm were shifted significantly and pore volume after adsorption of CG was reduced form 1.42 cm³/g to 0.276 cm³/g. The strong adsorption for CG on CTF_DCBP caused by mesorpore-filling mechanism due to size of the CG (size of CG 2.98 nm displayed in Figure S2) matched with the pore width of CTF_DCBP. It is note that the total pore volume MB, MO on CTF_DCBP also decreased from 1.42 cm³/g to 0.476 and 0.437. This indicated that the adsorption sites outside or inside the pores of CTF_DCBP were occupied efficiently by MB and MO, which was attributed to the adsorption mechanisms (cation-π, electrostatic interaction and hydrogen bonding) mentioned above.

The effects of pH for three dyes on CTF-1 (CTF_DCBP were not studied due to the similar pzc to CTF-1, only difference in pore structure) are presented in Figure 5. As shown in Figure 5, the K_d value increased gradually with the increasing pH for MB. MO and CG decreasd with the increasing pH, suggesting the same adsorptive interaction with CTF-1. CTF-1 is deprotonated and negatively charged with the increasing pH as the pzc around pH 7.2 (see SI Figure S6). Electrostatic interaction between cationic dye MB and CTF-1 gradually strengthened. However, the adsorption affinity could be suppressed when the pH continue increasing, because electrostatic repulsion was enhanced between CTF-1 and negatively charged MO and CG. In addtion, adsorption decreased slightly at pH range 7.2–10, indicating that higher pH favored hydrogen bonding interaction with the deprotonated triazine structure.

The effects of ionic strength for dyes on CTF-1 are listed in Figure 6. Clearly, ionic strength promoted MB adsorption on CTF-1. Salting-out effect is responsible for the increasing adsorption coefficient for MB (Grover & Ryall 2005). Similar result has been report in previous studies (Fontecha-Cámara et al. 2007; Eren & Afsin 2008). The adsorption of MO and CG decreased with the increasing ionic strength. MO and CG as anionic dyes, the electrostatic screening of the surface charge gradually enhanced with the increasing counterion species, leading to the lower adsorption coefficient. However, Ca²⁺ promoted the adsorption of CG on CTF-1, which may be caused by coordination or complexation between Ca²⁺ and CG.

Adsorption kinetics of MB and MO on CTF-1 and adsorption kinetics of CG on CTF_DCBP are presented in Figure 7. As show in Figure 7, small molecular dyes MB and MO adsorption on CTF-1 reached equilibrium within 3,500 and 5,000 min separately. Of note, adsorption equilibrium for CG on CTF_DCBP achieved equilibrium within 200 min. CTF-1 has narrow pore size distribution, and the dye molecules are not easy to diffuse into the interior, leading to relatively slow kinetics. The rapid kinetics for CTF_DCBP correlated closely with the shaped, open and wide pore size distribution, which is consistent with the results of adsorbent pore size distribution.

Simulation results of dyes adsorption on CTFs based on the pseudo-first-order kinetics and pseudo-second-order kinetics are presented in SI Figure S7, and the fitting parameters are listed in Table 2. As shown in SI Figure S7(b), the plot of t/q_t versus t presenteded a linear relation with a higher R² (0.997), suggesting that pseudo-second-order kinetic model could well depict the dyes' adsorption process to CTFs. In addition, the q_exp data (obtained from experimental data) are approximately identical to the q_cal data (calculated from the pseudo-second-order model) in Table 2, which further confirmed that the dyes' adsorption process on CTFs obeyed the pseudo-second-order kinetics model.

In this study, two different covalent triazine-based frameworks were synthesized and their adsorption for three dyes was systematically investigated. Characterization results indicated that the CTFs were successfully synthesized with a homogeneous and regular-shaped structure. Adsorption results demonstrated that the selected three dyes have strong adsorption on CTFs adsorbents because of cation–π, electrostatic interaction and hydrogen bonding interaction. Pore size distribution of CTFs is an important factor for adsorption particularly for bulky dyes CG. Acidic conditions are favorable for the adsorption of MB, while alkaline conditions are favorable for the adsorption of MO and CG. The results in this study indicate that CTFs are promising adsorbents for adsorptive removal of dyes in aqueous solution

This work was supported by Scientific Research Project of Nanjing Xiaozhuang University (2017NXY46), Excellent Science and Technology Innovation Group of Jiangsu Province, Innovative Practice of Environmental Engineering Subject based on New Engineering Construction.

All relevant data are included in the paper or its Supplementary Information.