Synergistic effect of graphitic-like carbon nitride and sulfur-based thiazole-linked organic polymer heterostructures for boosting the photocatalytic degradation of pharmaceuticals in water

Chemical Engineering Journal 494 (2024) 152843 Available online 4 June 2024 1385-8947/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Synergistic effect of graphitic-like carbon nitride and sulfur-based thiazole-linked organic polymer heterostructures for boosting the photocatalytic degradation of pharmaceuticals in water Raziyeh Zandipak a , c , * , Nader Bahramifar a , Morteza Torabi b , M´ onica Calero c , Mario J. Mu˜ nozBatista c , Rafael R. Solís c , * a Department of Environmental Science, Faculty of Natural Resources and Marine Sciences, Tarbiat Modares University, Noor, Iran b Department of Organic Chemistry, Faculty of Chemistry and Petroleum Sciences, Bu-Ali Sina University, Hamedan, Iran c Department of Chemical Engineering, Faculty of Sciences, University of Granada, Avda. Fuentenueva, 18014 Granada, Spain ARTICLE INFO Keywords: Covalent Organic Framework Graphitic Carbon Nitride Microwaved-assisted synthesis Photocatalysis Water Treatment ABSTRACT A sulfur-based COF has been combined with graphitic carbon nitride (CN) in microwave-assisted synthesis to build a COF-CN heterostructure with enhanced photocatalytic activity. The prepared COF-CN heterostructures were fully characterized, analyzing the textural (N 2 isotherms), structural (XRD and FTIR), chemical (elemental analysis and XPS), morphological (HR-STEM), optical (DRS-UV–Vis and photoluminescence) and electrochemical properties (EIS impedance, transient photocurrent, and flat band potential). Different COF-to-CN ratios (5–25 % of COF, wt.) were explored, defining a proportion (20 %) that led to optimum activity for the photocatalytic oxidation of organic contaminants of emerging concern (CECs) due to an enhanced separation of the photogenerated charges and lower bandgap value, 2.55 vs. 2.35 eV. The performance of the optimum COF-CN was further tested for other CECs, demonstrating its benefits compared to the bare CN. The materials displayed acceptable reusability and stability. The activation mechanism highlights the importance played by superoxide radicals and photogenerated holes. 1. Introduction Photocatalysis is a promising sustainable alternative to large-scale solar energy storage. Under convenient irradiation, a semiconductor can transform the absorbed photons into energy delivered to drive reactions that, if not thermodynamically impeded, the reaction rate under darkness is sparse [1]. The research of photocatalysis has skyrocketed during the last decades, exploring diverse applications such as the oxidation of aqueous pollutants either of inorganic or organic origin [2], water splitting [3], CO 2 photoreduction [1,4], N 2 fixation [5,6], or the selective synthesis of organic molecules with interest in the industry [7–10]. Among the abundant metallic semiconductors, in which TiO 2 outstands in almost all applications, the polymeric non-metallic graphitic-like carbon nitride has recently attracted attention due to its easy preparation from natural resources such as urea or easily synthesized as melamine. The bandgap of g-C 3 N 4 is lower than the benchmark TiO 2 , 2.7 vs 3.2 eV [11], which allows a better harvesting of the solar spectrum. However, like most photocatalysts, g-C 3 N 4 lacks an efficient separation of charges that minimizes the undesirable recombination effect. To overcome this limitation, g-C 3 N 4 has been modified, modifying the surface properties by the insertion of oxygenated groups [12], via non-metal doping [13–16], or combining it with other semiconductors to build a heterojunction [17–20]. The development of solids based on metal-organic frameworks (MOFs) [21,22] and covalent organic frameworks (COFs) [23–25] with photocatalytic properties, including solar light harvesting, has skyrocketed in the last years. Both are of paramount interest in the development of novel materials due to their textural properties, morphology at the nanoscale, and electronic properties, all of them easily tunable by a proper design of the scaffolding blocks and the synthesis at a molecular level. Compared to COFs, MOFs display much more morphologies due to the higher possibilities of multiple coordination around the metal cluster [26]. Nonetheless, COFs, built exclusively from covalent binding, lead to robust linkages that enable them to be stable in the presence of water avoiding hydrolysis [27], a recurrent drawback in most MOFs [28,29]. The high stability towards aqueous environments is due to the high * Corresponding authors. E-mail addresses: [email protected] (R. Zandipak), [email protected] (R.R. Solís). Contents lists available at ScienceDirect Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej https://doi.org/10.1016/j.cej.2024.152843 Received 1 March 2024; Received in revised form 1 June 2024; Accepted 3 June 2024 Chemical Engineering Journal 494 (2024) 152843 2 hydrophobicity of COFs [30]. Among all the plausible COFs being under research, covalent triazine framework polymers, prepared by the aromatic 1,3,5-triazine rings, were set up in 2008 [31] and have acquired utmost importance due to the high chemical stability of the frameworks conjugated structure prepared from this unit. Moreover, the COFs based on triazine are sensitive to radiation and display properties as semiconductors [32]. The condensation of tri(4-formylphenoxy) cyanurate (TFPC) via dynamic nucleophilic aromatic substitution has been reported recently, obtaining a tacked planar material with good stability and high CO 2 adsorption capacity [33]. COF bearing a π -electron deficient triazine core and a π -electron rich core containing phenyl and naphthalene rings connected with –O and –N donor Lewis basic sites in hexagonal honeycomb layers has been reported from the imine reaction of TFPC and naphthalene diamino [34]. This work reports a sulfur-based COF made from the polycondensation of TFPC and naphthalene diamine. The resultant COF displayed visible radiation absorption, i.e. greenish appearance, but was highly hydrophobic. To overcome the hydrophobic drawback, COFs are commonly modified, joining them to other semiconductors with favorable surface interaction with water [35]. For that reason, the resulting COF was combined with graphitic-carbon nitride, a metal-free polymeric semiconductor [11,36,37], in a microwave-assisted synthesis pot, considerably reducing the time of synthesis [38,39]. The prepared COFCN heterostructure displayed enhanced photocatalytic properties for the degradation of contaminants of emerging concern in water. 2. Experimental section 2.1. Synthesis of COF and COF-CN heterostructures 2.1.1. Synthesis of the COF and graphitic-like carbon nitride (CN) For the synthesis of the COF, firstly tri-(4-formylphenoxy) cyanurate (TFPC) was obtained from the reaction of cyanuric chloride with 4hydroxybenzaldehyde. For that, 20 mmol of 4-hydroxybenzaldehyde and 20 mmol of NaOH were dissolved in 80 mL of deionized water. After, 4 mmol of cyanuric chloride, dissolved in 40 mL of acetone, was slowly dropwise added to the previous solution. The mixture was stirred at room temperature for 5 h to complete the reaction, leading to the formation of a white solid which was filtered and washed with ethanol and aqueous sodium bicarbonate solution. The COF was prepared by adding 0.456 mmol of TFPC, 0.684 mmol of 1,5-diaminonaphthalene, 2 mmol of sulfur, 0.2 mL of dimethyl sulfoxide (DMSO), and 5 mL of dimethyl formamide (DMF). The resulting solution was submitted to thermal treatment under reflux at 140 ◦C for 24 h to promote the polymerization reaction, see Fig. 1A. The resulting greenish precipitate was washed several times with methanol, tetrahydrofuran, and trichloro-ethylene to remove the unreacted chemicals. The washed solid was dried under a vacuum at 80 ◦C overnight. The graphitic-like carbon nitride sample (CN) was prepared by thermal polymerization of melamine at 550 ◦C for 1 h under an air atmosphere in a sealed crucible. The yellowish solid was treated with HCl 1 M (7 g/L of g-C 3 N 4 ) to promote delamination under sonication and washed several times with water. 2.1.2. Synthesis of the COF-CN heterostructures The COF-CN heterostructures were prepared by microwave-assisted solvothermal treatment, see a scheme of the synthesis route in Fig. 1B. For that, 1 g of CN and a defined amount of COF were placed in a Teflon line vessel with 25 mL of DMF. The vessel was sealed and magnetically stirred inside a high-pressure multimode flexiWAVE platform for microwave synthesis of Milestone®. The temperature was raised by the action of microwave radiation to 120 ◦C in 15 min and held at this value for 30 min. After, the samples were submitted to natural cooling. The resulting COF-CN heterostructures were washed several times with ultrapure water, and the recovered solid by centrifugation was dried overnight at 105 ◦C. The resulting samples were labeled as x%COF-CN were x (5–25) means the mass percentage of the COF placed in the synthesis procedure. For comparison purposes, the CN sample obtained by melamine polymerization was also submitted to microwave treatment following the same steps but lacking the addition of the COF. 2.2. Characterization of the COF-CN heterostructures The crystallinity of the samples was studied by X-Ray Diffraction (XRD), in a Bruker D8 Discover (50 kV, 1 mA) diffractometer working with X-Ray from the Cu K α (λ =1.5406 Å) equipped with a Pilatus3R 100 K-A detector. The diffractograms were recorded at room temperature in the 2θ range of 4-70◦, under a rate of 30 s per 0.02◦. The software QualX® was used for the interpretation of the diffractograms and the determination of the crystal size through Scherrer’s equation. The crystallite size was estimated by Scherrer’s equation from the most intense peak and the interlayer spacing of the graphitic-like carbon nitride was estimated from the (002) peak. The relation L crystal /d layer was used as an estimation of the number of layers of the graphitic-like structure [40]. The structural properties were also assessed by Fourier Transform InfraRed (FTIR) spectroscopy equipped with Attenuated Total Reflectance (ATR). The FTIR spectra were recorded in a Spectrum65 device Fig. 1. Reaction scheme for the synthesis of the sulfur-based COF (A) and synthesis preparation of the COF-CN heterostructures by microwaved-assisted method (B). R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 3 from Perkin-Elmer®, monitoring the signal in the range 650–4000 cm −1 . The identification of the peaks with the plausible functional groups was carried out with the assistance of irAnalyze-RAManalyze software from LabCognition GmbH. In addition, Raman spectroscopy was conducted in a Micro-Raman Dispersive JASCO NRS-5100 spectrometer equipped with a diode laser of excitation of 785 nm (500 mW). The textural properties were evaluated by N 2 physisorption. The N 2 adsorption–desorption isotherms at 77 K were obtained in a Sync 220 device of 3P Instruments®. The samples were previously degassed at 150 ◦C under vacuum overnight in a Prep J4 degasser station of 3P Instruments®. The elemental composition was determined in a TrueSpec® Micro CHNS analyzer from Leco®. Moreover, the surface chemistry was analyzed by X-ray Photoelectron Spectroscopy (XPS) in a Kratos AXIS UltraDLD device equipped with an X-ray source from Al K α . The spectra were referenced to adventitious C 1s , ascribed to 284.6 eV. The deconvolution of the spectra was conducted using the software XPSpeak 4.1®, applying a Shirley-type correction of the baseline background. The morphology and element distribution of the COF-CN nanoparticles was observed by High-Resolution Scanning Transmission Electron Microscopy (HR-STEM) imaging in a Thermo Scientific TM Talos TM F200X (200 kV) equipped with High-Angle Annular Dark-Field imaging (HAADF) detector and Energy Dispersive X-Ray spectroscopy (EDX) microanalysis for the analysis of the composition and study of the element distribution. The photoluminescence (PL) technique was conducted as an indirect measure of the recombination rate of the photogenerated electron-hole pair [41,42]. The analysis was conducted in a Varian Cary Eclipse fluorescence spectrometer, setting 365 nm as the excitation wavelength (slit 2.5 nm) and registering the emission spectra (slit 2.5 nm) from 400 to 600 nm. The electrochemical characterization of the samples in terms of photocurrent, electrochemical impedance spectroscopy (EIS), MottSchottky analysis, and transient photocurrent tests, was carried out in a conventional three-electrode cell with 0.2 M Na 2 SO 4 (pH =7.0) as the electrolyte solution. The working electrode was prepared on indiumdoped tin oxide (ITO) transparent conductive glass, a Pt wire and a saturated calomel electrode (SCE) were used as the counter and reference electrodes, respectively. The measurements were performed using a CHI 660C Electrochemical Workstation (USA). 2.3. Photocatalytic tests The photocatalytic activity of the COF-CN heterostructures was assessed in the degradation of acetaminophen (ACE) in water. This contaminant was selected for a preliminary evaluation of the different COF-CN ratios based on its occurrence in different effluents including wastewater [43,44]. The photoreactor used was equipped with two UVA lamps of 9 W each emitting at a maximum wavelength of 365 nm. The lamps were placed in the inner space of a jacketed annular reactor made of borosilicate glass. The aqueous solution with the COF-CN sample was pumped to the inner space, magnetically stirred, from an auxiliary tank, also a stirrer, and equipped with refrigeration to maintain the temperature at 20 ◦C. Detailed information about the dimensions of the photoreactor can be checked in a previous work [19]. Air was bubbled into the auxiliary tank to ensure the saturation of O 2 in the aqueous solution. The photocatalytic experiments started feeding the system 350 mL of solution. The photocatalyst was next added to set a dose of 0.5 g L -1 . An adsorption period of 30 min in the absence of radiation, i.e. darkness, was carried out to ensure that the adsorption equilibrium was reached. Next, the lamps were switched on, and the evolution of the concentration of acetaminophen was monitored. The photocatalyst was removed by filtration with syringe filters (Millex PVDF, 0.45 μ m). The photocatalytic tests involved experiments of degradation of acetaminophen at initially 5 mg L -1 for selecting the best performance. For further testing, the optimum photocatalyst was tested in the degradation of a mixture of compounds, i.e. a mixture of acetaminophen (ACE), caffeine (CAF), antipyrine (ANT), ciprofloxacin (CIP), sulfamethoxazole (SMX) and diclofenac (DCF), initially at 2 mg L -1 each. These compounds were considered based on their occurrence in wastewater [45]. Photolysis tests, i.e. absence of photocatalyst, were carried out for comparison purposes in both cases. The concentration of acetaminophen was analyzed by high-pressure liquid chromatography (HPLC) in a Water Alliance e2695 HPLC device, equipped with a 2998 photodiode Array (PDA) detector. A Zorbax Bonus-RP column (5 μ m, 4.6x150 mm) was used as the stationary phase. The mobile phase, pumped at a flow rate of 1 mL min −1 under the isocratic mode, was a mixture of acetonitrile (A) and ultrapure water acidified with 0.1 % (v/v) of trifluoroacetic acid (B). The analysis of ACE was carried out with an A: B mixture of 60:40 (vol.), and the quantification at 240 nm. For the analysis of the mixture, a gradient program was set up, starting with 5:95 pumped at 1 mL min −1 during 5 min, raising the acetonitrile proportion to reach 95:5 in 30 min and further kept at this proportion for 5 min. The quantification of ACE and ANT was conducted at 240 nm, CAF and SMX at 270 nm, DCF at 275 nm, and CIP at 280 nm. The influence of pH in the range 3–9 on the photocatalytic activity of the most active sample was tested by adjusting the pH with NaOH or HCl 0.1 M. The effect of the presence of the most common inorganic anions on the photocatalytic performance of the most active sample was conducted by adding Cl – (2.50 mM), NO 3 – (1 mM), or HCO 3 – (3 mM) at the maximum values in wastewater after reclamation treatment [46]. The role played by the different reactive oxidation species was assessed by tests in the presence of chemical scavengers, which were added to the solution before loading in the photoreactor. The role played by the superoxide radical was evaluated by replacing air with N 2 , adding 1 mM p-benzoquinone (p-BZQ) or 1 mM disodium 4,5-dihydroxybenzene-1,3-disulfonate (tiron). 10 mM tert-butyl alcohol (TBA) was added to scavenge the contribution of hydroxyl radicals. The importance of the photogenerated holes was assessed by adding 10 mM oxalic acid. Moreover, the impact of HO • was additionally assessed by an indirect probe method based on terephthalic acid (TPA) [47,48]. The nonfluorescent terephthalic acid traps HO • triggering the formation of the fluorescent 2-hydroxy-terephthalic acid (2-HO-TPA) [49]. Thus, the monitoring of 2-HO-TPA formation can be considered as an indirect indicator of the HO • presence in the aqueous solution. For this reason, a test with TPA 1 mM was carried out [50]. The analysis of 2-HO-TPA with the reaction time was conducted in a Varian Cary fluorescence spectrometer setting the excitation wavelength at 315 nm (slit 2.5 nm) and the emission spectra recorded between 360–600 nm (slit 2.5 nm), registering the maximum of the 2-HO-TPA spectrum peak at 420 nm. The correlation of the peak high with the concentration of the standard 2-HO-TPA was carried out in the range 0.5–5 μ M, leading to a limit of detection of 0.21 μ M. 3. Results and discussion 3.1. Characterization of the COF-CN heterostructures The crystalline structure of the COF-CN heterostructures was analyzed by X-ray diffraction. Fig. 2A illustrates the diffractograms of the prepared materials. The diffractogram of the graphitic-like carbon nitride (CN) displayed diverse crystalline peaks, from which a sharp peak outstands the rest. The condensation of the precursor, i.e. melamine, triggers the formation of tris-s-triazine units which are polymerized in plates interacting by the π - π * interactions leading to a certain number of layers aggrupation [51,52]. Consequently, an intense peak located at roughly 27◦is defined in CN materials, associated with the (002) diffraction plane defined by the interplanar aromatic interaction from π - π * interactions of the aromatic heptazine rings [53]. The polymerization conditions such as the temperature strongly impact the sharpness of this peak. The number of layers at condensation R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 4 temperatures of 500 and 600 ◦C has been reported between approximately 22 and 31◦, respectively [54]. A tentative estimation of the number the layers from the division of the crystal size and the d-spacing of the XRD interaction [40] suggested roughly 40 layers for the CN sample. The diffractogram of CN defines a second peak, much less intense and considerably broadened, placed at ca. 13◦. This peak is attributed to the (100) plane, distinct from the intralayer spacing of the heptazine rings [55]. In the case of the COF, a much less intense diffraction pattern was obtained, which suggests the formation of a moderate crystalline polymer, which follows the already reported polymerization of TPFC and naphthalene diamino [34]. Similarly, a hexagonal staked layered structure with different ring widths due to sulfur incorporation is tentatively envisaged. The COF-CN structures defined a strong peak from the interlayer stacking of CN nanosheets, predominant from the COF XRD pattern due to the higher proportion and crystallinity of CN if compared to the COF. As reported in other gC 3 N 4 heterostructures, the modification of the material and the combination with other semiconductors alters the graphitic structure leading to less crystalline materials [56]. There was no observed tendency regarding the crystal size of the heterostructures, providing an estimation of the number of layers for the graphitic structure of 40–45. The structural properties were analyzed by FTIR. The spectra depicted in Fig. 2B demonstrate important changes in their FTIR footprints. A broadband registered at 3000–3300 cm −1 appeared in CN due to the terminal N–H and O–H stretching, attributed to the primary amine (–NH 2 ) and adsorbed water molecules [42]. The decrease of the terminal groups may be associated with the interaction with the COF structure. The heating through microwave radiation assists in the creation of new bonds through a hot-spot mechanism [57]. Moreover, the interaction of COFs with 2D materials such as g-C 3 N 4 takes place by π - π interactions [58] which seems not to be the case since a 2D COF cannot be surmised from the microscopy analysis. The vibration of the aromatic tri-s-triazine rings leads to different peaks. Thus, the vibration of secondary nitrogen (N 2 C) and tertiary nitrogen (N 3 C) define peaks located at 1225 and 1312 cm −1 , respectively [43]. In addition, the peaks placed at approximately 1450, 1535, and 1620 cm −1 can be identified as the aromatic CN vibration [36,44,45]. The peak appearing at 810 cm −1 has been associated with the out-plane vibration of the heptazine rings [43]. The sulfur-based COF led to a very different FTIR pattern. The wide band defined in the CN due to terminal −NH x and adsorbed water molecules was not registered in the COF, suggesting a complete condensation of the naphthalene diamino groups during the COF synthesis. The presence of the triazine group in the COF led to some vibrational peaks such as C =N at 1570 and 1504 cm −1 , C-N stretch at 1357 cm −1 , C-H bend at 780 cm −1 , and C-O-C stretch leading to peaks at 1294 and 1155 cm −1 . The peak placed at 625 cm −1 may be attributed to C-S-C vibration. To further investigate the structural properties of the samples, Raman spectroscopy was conducted, see spectra in Fig. 2C. The sample CN described the typical reported Raman spectrum of graphitic carbon nitride. The Raman peaks located at 707 cm −1 and 979 cm −1 are attributed to the breathing modes of the s-triazine ring which describe the out-of-plane deformation vibration between the layer of the g-C 3 N 4 heterocycles [59]. The peak at 480 cm −1 is associated with the twisting vibration of the heptazine ring [60]. The peak at 1227 cm −1 can be ascribed to the double N =C (sp 2 ) bending vibration.[60] However, those samples modified with the insertion of the COF led to the absence of Raman peaks, as shown for 20 %COF-CN. The textural properties were assessed by N 2 adsorption–desorption isotherms conducted at −196 ◦C. The physisorption isotherms are depicted in Fig. 3 and the textural properties are summarized in Table 1. A typical type IV pattern behavior with an H3 hysteresis loop [61–63] is described by the CN sample, as compatible with a mesoporous solid Fig. 2. XRD diffractograms (A), FTIR spectra (B), and Raman spectra (C) of the COF-CN heterostructures. R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 5 composed of plate-like aggregates [64]. The BET area was as low as 7.7 m 2 /g and poor pore volume, 0.034 cm 3 g −1 , was found, typical values reported for graphitic carbon nitride prepared from the polymerization of melamine [65]. The COF displayed a higher specific area, 35.2 m 2 /g , and an increased pore volume, i.e. 0.191 cm 3 g −1 , defined its totality as mesopore. The COF-CN heterostructures performed textural values in between the obtained by CN and COF. The elemental composition of the samples is available in Table 1. The C:N (at.) ratio of CN was 1.5, slightly higher than the theoretically expected for graphitic-like carbon nitride, but very similar to the value reported in other works using melamine as the precursor [8]. This ratio was gradually reduced as the proportion of the COF was raised. The surface chemical environment was analyzed by XPS. The samples CN, COF, and 20 %COF-CN were selected for a detailed analysis and their XPS spectra after deconvolution are depicted in Fig. 4. The C 1s region of graphitic-like carbon nitride is commonly deconvoluted in three main contributions, i.e. a peak of sp 2 bonds linked to N-C=N bonds (288 eV), a peak of sp 3 carbon in C-C/C-N bonds (285.5–286.0 eV), and a peak of sp 2 carbon bonds present in aromatic C-C/C=C (~284.4 eV) [8,66]. According to the C 1s spectrum of CN, the contribution of N-C = N, representative of the triazine rings outstands from the rest, as reported in the literature for carbon nitrides prepared from melamine as a precursor [7,8]. In the case of the COF, the aromatic C-C/C=C bonds displayed more importance in the C 1s spectrum, due to the presence of the aromatic rings of naphthalene rings. The sp 3 contribution observed Fig. 3. N 2 adsorption–desorption isotherms of the COF-CN heterostructures. Table 1 Crystal and textural properties, elemental composition, and optical properties of CN-COF heterostructures. Sample L crystal (nm) d (Å) n S BET (m 2 g −1 ) V T (cm 3 g −1 ) C EA (wt. %) N EA (wt. %) S EA (wt.) N/C (at.) E BG (eV) CN 12.93 3.248 40 7.7 0.034 34.2 59.7 0.0 1.50 2.55 5 %COF-CN 15.81 3.250 47 19.2 0.082 36.0 59.0 0.0 1.40 2.55 10 %COF-CN 11.86 3.245 37 23.4 0.101 38.5 56.4 0.2 1.26 2.50 20 %COF-CN 12.20 3.252 38 30.1 0.102 39.8 54.4 1.0 1.17 2.35 25 %COF-CN 14.73 3.238 45 16.4 0.069 37.1 54.5 1.9 1.26 2.35 COF 70.35 4.126 −35.2 0.191 46.0 39.0 7.4 −2.20 L crystal , crystallite size from the highest peak by Scherrer’s equation; d, interlayer spacing from the highest peak; n, number of layers; S BET , specific total surface area obtained from the BET method; V T, total pore volume obtained from the N 2 uptake at p/p 0 ~ 0.99; C, N, and S composition from elemental analysis (EA); and E BG , bandgap energy estimated from the Tauc plot method. Fig. 4. High-resolution XPS spectra of C, N, and S regions for CN, COF, and 20 %COF-CN samples. R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 6 in the COF could be associated with the ether C-O-C bonds [67,68], much less intense than the aromatic sp 2 C-C contribution. A new peak due to the sulfur incorporation as C-S was added in the deconvolution of the C 1s region, placed at roughly 285.0 eV [26,69], leading to a peak with a similar area as the sp 3 C-O-C ether. Regarding the N 1s region, in the case of graphitic carbon nitride materials, three contributions can be ascribed, i.e. tertiary N from (C) 3 –N rings (N 3C , 399.9 eV), sp 2 -bonded in the form of N–C=N (N 2C , 398.4 eV), and terminal −NH x groups (400.8 eV) [54,70–73]. The predominant contribution observed in the CN sample was N 2C associated with the aromatic triazine rings, followed in importance by N 3 C and terminal −NH x , a behavior very similar to the already reported for this type of material [7,8]. The N 1s spectrum of the COF did not define amino −NH x groups and mostly N 2 C, which could be attributed to the triazine units of the TFPC precursor of the COF. Finally, pertaining to the analysis of the S 2p region, a major contribution of thioether C-S-C was registered, with binding energies of 163.7 eV (2 p3/2 ) and 164.9 eV (2 p1/2 ) [69,74–77]. A minor contribution of sulfate was also registered, binding energies of 169.0 eV (2 p3/2 ) and 170.0 eV (2 p1/2 ) [78]. The heterostructure 20 %COF-CN displayed a contribution of both the CN and COF in the contributions of the above-mentioned regions, leading to a greater contribution of those provided by the CN due to the higher proportion if compared to the COF. The morphology of some selected samples, i.e. CN, COF, and the 20 %COF-CN heterostructure, was studied by HR-STEM. The composition and distribution of the elements were also scanned by EDX mapping. Fig. 5 portrays some of the pictures obtained. The imaging of CN led to nanoparticles of different sizes up to 1 µm. In general, CN is characterized by the agglomeration of parallel sheets, in which a homogeneous distribution of C and N was appreciated. There was no obtained appreciable presence of oxygen in EDX analysis. The COF imaging depicted bigger aggregates of several micrometers. High-resolution imaging of these aggregates allowed us to define them as conglomerates of rod-shaped particles (see Fig. 5E). The heterostructure 20 %COFCN led aggregates of similar size and shape as the COF. The EDX mapping draws a suitable picture of the heterogeneity of the sample. As the CN particles are richer in nitrogen than carbon, their presence in the heterostructure is enlightened by N-bright particles, attributable to laminar sheets of graphitic carbon nitride, as illustrated in the subfigures J-L of Fig. 5. The accumulation could be associated with the COF as the distribution of S suggests. Alternatively, the identification of the CN and COF areas in the heterostructure is demonstrated in the EDX spectrum, portrayed in subfigures M and N of Fig. 5, respectively. The presence of the COF helped to wrap CN particles in bigger aggregates, reaching an intimate contact between both components. The optical properties were characterized by employing the DRSUV–visible technique, leading to the absorption spectra depicted in Fig. 6A. As appreciated, the CN sample performed high absorbance in the UV region to dramatically decay at roughly 400 nm, describing an absorption peak around 300–400 nm. This peak can be attributed to the π – π * electronic transitions in the conjugated ring systems heptazine units [79], conferring to the material a yellowish aspect. The sulfurbased COF, of greenish aspect, displayed a strongly red-shifted spectrum beyond 800 nm. These results provide evidence of the boosted light-harvesting capability of the COF in a wide range of visible light region due to the delocalized π electrons in the polymeric structure, which is common in other triazine-based COFs [80]. The COF displayed two maximum absorption peaks, one placed at 300–400 nm similar to the CN sample and other broader at 500–800 nm, responsible of the greenish color of the material. The COF-CN heterojunctions displayed intermediate spectra, resembling the CN and COF depending on their relative proportion, being a raised amount of COF in the heterojunction positive for harvesting the radiation in the visible region. The bandgap values were determined by applying the Tauc plot method, see Fig. 6B. The COF displayed a lower band gap if compared to CN, 2.20 vs 2.35 eV. The 2.30 eV obtained for CN is lower than the typical 2.70 eV reported in the literature, probably due to the thermal treatment undergone in the microwave synthesis. The COF-CN heterostructures displayed intermediate bandgap values, as shown in Table 1, playing the COF with a decreasing effect. 3.2. Photocatalytic activity of the COF-CN heterostructures The photocatalytic activity of the COF-CN heterostructures was tested for the degradation of acetaminophen. Fig. 7A depicts the results Fig. 5. HR-STEM images and EDX mapping of CN (A-D), COF (E-H), and 20 %COF-CN (I-L) and EDX spectra of selected areas of 20 %-COF-CN (M and N). R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 7 achieved. In the absence of radiation, ACE was slightly photolyzed, barely 38 %. The photocatalytic activity of carbon nitride led to roughly 90 % of the removal of ACE in 3 h, with a pseudo-first order rate constant of 1.00 h −1 , see Fig. 7B. Although the activity of the COF was tested, it led to insignificant degradation of ACE due to a lack of suspension of the COF in the aqueous sample. The high hydrophobicity of the COF forced the particles to be concentrated on the surface of the solution, discharging the possibility of its use without any modification in the structure. However, the incorporation of the COF in the CN structure enhanced the kinetics if compared to bare CN, due to the enhanced delocalization of the photogenerated charges, as proved in the reduction of the photoluminescence peak registered in the COF-CN heterostructures. The rise of COF ratio exerted a beneficial effect on the pseudo-first order rate constant, until reaching a maximum value of the k =2.28 h −1 with the sample 20 %COF-CN, which supposes an overdoubled value concerning the CN sample. An excess of COF over 20 % worsened the results, as an excess of COF results in an excess of recombination of the electron-hole pair after excitation, as the photoluminescence spectra depicted in Fig. 7C of the sample 25 %COF-CN suggests. Furthermore, the evolution of the PL peak, whose maximum is defined at ca. 460 nm, can be correlated with the photocatalytic activity, leading to an inverse relationship between the pseudo-first order rate constant and the intensity of the PL peak, see Fig. 7B. The kinetics performance obtained with the 20 %COF-CN sample was compared to the reported by similar materials in the literature, see Table 2. The tentative comparison should be taken with caution since the attained results considerably depend on the experimental conditions, i.e. nature of the radiation source and intensity. Notwithstanding the experimental conditions, the pseudo-first order rate constant k ACE = 2.28 h −1 is within the same order of magnitude as other graphitic carbon nitride-based materials combined with TiO 2 , or metal–organic frameworks. Other reported works have stated abnormally higher values that Fig. 6. DRS-UV–visible spectra (A) and determination of the band gap by the Tauc plot method (B) of the COF-CN heterostructures. Fig. 7. Photocatalytic degradation of ACE with COF-CN heterostructures. Influence of the COF ratio on the temporal evolution of the normalized concentration (A), the pseudo-first order rate constant (B), and photoluminescence spectra (C). Experimental conditions: V =250 mL; T =20 ◦C; C COF-CN =0.5 g/L; C ACE,0 =5 mg/L. R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 8 can be attributed to more favorable radiation conditions, or the presence of highly active metallic species. The materials with the slowest kinetics were the metal–organic frameworks without combination with other semiconductors. Regarding covalent organic frameworks, there is scarce information about the photocatalytic degradation of contaminants of emerging concern; however, some of the reported k ACE values are like those obtained with the synthesized material of this work. The photoelectrochemical properties of the prepared COF-CN heterostructures were further characterized to reveal if the reason why the sample 20 %COF-CN outstands the rest is due to better electronic mobility, as tentatively suggested by the photoluminescence technique. The migration of charges, a crucial role during photocatalytic activation, was assessed by EIS characterization. Bearing in mind that the smaller radius curvature of the Nyquist plot means a higher mobility rate [48,92], as pictured in Fig. 8A, the addition of COF positively impacted a better electronic migration, resulting in lesser charge transfer resistance, which contributes to enhancing the charge migration. Concretely, the order of the radius depicted in the Nyquist diagram was 20 %COF-CN < 25 %COF-CN <10 %COF-CN <5 %COF-CN <CN. The lowest radius performed by 20 %COF-CN demonstrates the minimum resistance to electronic mobility and therefore the optimum ratio of COF added to the graphitic-like carbon nitride. Fig. 8B depicts the transient photocurrent response of the COF-CN samples for four on–off cycles under UVA radiation. It is well stated that the higher the photocurrent response, the higher the separation efficiency of the photo-generates charges [92]. The higher photocurrent intensity is also interpreted as a faster photoexcited electron transferred to the surface and less recombination of photoexcited electron-hole pair [93]. The order of photocurrent responses follows the results attained in the EIS analysis, since under illumination the photocurrent density was 20 %COF-CN >25 %COF-CN >10 %COF-CN >5 %COF-CN >CN. It can be concluded, therefore, that the optimum COF ratio in the sample 20 %COF-CN displayed the lowest resistance to electronic mobility, the highest density of photogenerated charges under illumination, and the lowest recombination rate. Fig. 9A illustrates the effect of the pH on the pseudo-first rate constant of ACE degradation (k ACE ) with 20 %COF-CN. As portrayed, the k ACE values gradually augmented to reach the maximum value at circumneutral conditions and decreased at alkaline conditions. The ACE molecule displays a pKa =9.5 [94], which means that the molecule is neutral in all the range of pH studied, therefore the electrostatic interactions do not pose relevant importance on the mechanism. At alkaline conditions, it has been stated a negative influence is due to the inhibition played by less effective migration of photogenerated charges at pH >8 [95,96]. The 20 %COF-CN performs the best kinetics at close to neutral or slightly acidic conditions. The presence of inorganic species such as Cl – , NO 3 – , and HCO 3 – has been claimed to negatively interfere with the photocatalytic activity of Table 2 Comparison of the photocatalytic performance of acetaminophen degradation with similar materials. Material Radiation conditions C ACE,0 (mg/ L) C catalyst (g/L) k ACE (h −1 ) Ref. g-C 3 N 4 based materials TiO 2 /g-C 3 N 4 UVA lamp (9 W) 365 nm 10.0 0.2 2.18 [81] WO 3 /g-C 3 N 4 Xe lamp (300 W) >400 nm 10.0 1.0 7.80 [82] CeO 2 /I,K-g-C3N4 Visible (64 W) 465 nm 10.0 2.0 2.34 [83] g-C 3 N 4 /UiO-66-NH 2 UVA lamps (18 W) 365 nm 5.0 0.5 2.00 [19] g-C 3 N 4 /NH 2 -MIL-101 (Fe) Solar light (+H 2 O 2 ) 20.0 0.4 6.90 [84] Covalent Organic Frameworks materials COF based on phenylenediamine with 2 ortho heterocyclic N Visible (125 W) >400 nm 5.0 0.3 7.68 [85] Ketonenamine-based COF/ AgI Xe lamp (300 W) 5.0 0.3 2.79 [86] Sulfur-based thiazolelinked COF/ g-C 3 N 4 UVA (18 W) 365 nm 5.0 0.5 2.28 This work Metal Organic Frameworks materials UiO-66-NH 2 Xe lamp (600 W) >320 nm 5.0 0.25 0.37 [87] ZIF-8 Xe lamp (500 W) >400 nm 1.0 0.5 0.14 [88] NH 2 -MIL-125(Ti-Zr) Xe lamp (600 W) >290 nm 5.0 0.25 0.73 [89] Pd/NH 2 -MIL-125 Xe lamp (600 W) >290 nm 5.0 0.25 0.95 [90] FeCo-MOF Sunlight 20.0 0.5 1.86 [91] Fig. 8. Nyquist plot of electrochemical impedance spectroscopy (A) and transient photocurrent response (B) of COF-CN heterostructures. R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 9 photocatalysts due to their interaction with the generated reactive oxygen species. Accordingly, individual ACE degradation assays were conducted with 20 %COF-CN in the presence of the above-mentioned anions at their environmentally relevant concentration. The addition of Cl – 2.5 mM displayed an unimportant effect whereas the presence of NO 3 – 1 mM exhibited a slight deceleration of the kinetics of ACE degradation, see the pseudo-first order rate constant variation in Fig. 9B. The presence of these anions in a photocatalytic process can lead to the formation of the less reactive species Cl • and NO 3 • , with the consequent quenching effect on the target contaminant degradation [97]. In the test carried out in the presence of HCO 3 − , the greatest decrease of the pseudofirst order rate constant was registered due to the combined effect of the less favorable pH, since the presence of HCO 3 – raised the pH to 8.5–9.0 and the formation of CO 3 •− radicals, which have demonstrated to possess lower reactivity, which often is considered as one of the main disadvantages for advanced oxidation technologies in practice [98]. Based on the boosted activity of the heterostructure 20 %COF-CN, additional tests were carried out to assess the behavior in the degradation of other contaminants of emerging concern (CEC). Fig. 10 shows the results attained during the degradation of a mixture of six CECs, i.e. acetaminophen, caffeine, antipyrine, ciprofloxacin, sulfamethoxazole, and diclofenac. For comparison purposes, a photolysis and the performance of the CN sample were also accomplished. Fig. 10A depicts the temporal evolution of the CEC concentration during the performance of photocatalytic degradation with 20 %COF-CN. As illustrated, ACE, CIP, and DCF were the most reactive, leading to a complete abatement in 120 min. In the case of DCF, a considerable adsorption contribution was registered in the previous 30 min of darkness step. DCF has been reported to be adsorbed onto carbon nitride [99], reaching mesoporous formulas with a maximum DCF uptake of up to 122 mg g −1 [100]. The other three contaminants, e.g. CAF, SMX, and ANT, showed more resistance to degradation. The recalcitrance of CAF [101] and ANT [102] towards photocatalytic oxidation has been widely reported in the literature. Interestingly, it is observed in their temporal profiles a slag period that coincides with the degradation of the three previously mentioned. Once a considerable removal, i.e. roughly 90 min of the most reactive is reached, the slope of the most recalcitrant increases (SMX, CAF, and ANT), defining kinetics adjustable to a pseudo-first order. Fig. 10B illustrates the pseudo-first order rate constant of photolysis, and photocatalytic oxidation with CN and 20 %COF-CN. From these results, it is also evidenced the easiness of the degradation of CIP, since it is simply removed by the sole action of radiation. CIP displays an important absorption activity between 300–350 nm [103,104], which makes it greatly sensitive for being photolyzed at 365 nm [105]. Excepting CIP, the rest of the CECs presented quite limited removal by photolysis. The photocatalytic oxidation of CN enhanced led to the removal of the CECs with a reactivity order DCF >CIP >ACE ≫ SMX >CAF >ANT. The Fig. 9. Photocatalytic degradation of ACE with the 20 %COF-CN heterostructure. Influence of the pH (A) and the presence of anions (B) on the pseudo-first order rate constant. Experimental conditions: V =250 mL; T =20 ◦C; C 20%COF-CN =0.5 g/L; C ACE,0 =5 mg/L; pH =3–9; C Cl- =2.5 mM; C NO3– =1 mM; C HCO3– =3 mM. Fig. 10. Photocatalytic degradation of a mixture of six CECs with 20 %COF-CN heterostructure. Temporal evolution of the normalized concentration under photocatalytic degradation with 20 %COF-CN (A) and the pseudo-first order rate constant of photolysis, photocatalytic degradation with CN or 20 %COF-CN (B). Experimental conditions: V =250 mL; T =20 ◦C; C CATALYST =0.5 g/L (if necessary); C CEC,0 =2 mg/L (each). R. Zandipak et al. Chemical Engineering Journal 494 (2024) 152843 16 nanoparticles for the treatment of bisphenol A in wastewater: Synthesis, photocatalytic activity, degradation byproducts and reaction pathways, Appl Catal B 241 (2019) 598–611, https://doi.org/10.1016/J.APCATB.2018.09.039. [118] S. Gligorovski, R. Strekowski, S. Barbati, D. Vione, Environmental implications of hydroxyl radicals ( • OH), Chem Rev 115 (2015) 13051–13092, https://doi.org/ 10.1021/CR500310B. [119] M. Sahni, B.R. Locke, Quantification of hydroxyl radicals produced in aqueous phase pulsed electrical discharge reactors, Ind Eng Chem Res 45 (2006) 5819–5825, https://doi.org/10.1021/IE0601504. [120] L.L.S. Silva, W. Abdelraheem, M.N. Nadagouda, A.M. Rocco, D.D. Dionysiou, F. V. Fonseca, C.P. Borges, Novel microwave-driven synthesis of hydrophilic polyvinylidene fluoride/polyacrylic acid (PVDF/PAA) membranes and decoration with nano zero-valent-iron (nZVI) for water treatment applications, J Memb Sci 620 (2021) 118817, https://doi.org/10.1016/J.MEMSCI.2020.118817. [121] L. Wang, B. Li, D.D. Dionysiou, B. Chen, J. Yang, J. Li, Overlooked formation of H 2 O 2 during the hydroxyl radical-scavenging process when using alcohols as scavengers, Environ Sci Technol 56 (2022) 3386–3396, https://doi.org/10.1021/ ACS.EST.1C03796. [122] A. Krishnan, S. Beena, M. Chandran, Fabrication and evaluation of CeO 2 -Fe 2 O 3 mixed oxide for hydrogen evolution by photo water splitting reaction under visible light irradiation, Mater Today Proc 18 (2019) 4968–4976, https://doi.org/ 10.1016/J.MATPR.2019.07.489. [123] B.R. Shah, U.D. Patel, Mechanistic aspects of photocatalytic degradation of Lindane by TiO 2 in the presence of Oxalic acid and EDTA as hole-scavengers, J Environ Chem Eng 9 (2021) 105458, https://doi.org/10.1016/J. JECE.2021.105458. [124] B.G. Ershov, E. Janata, M.S. Alam, A.V. Gordeev, A pulse radiolysis study of the reactions of the hydrated electron and hydroxyl radical with the oxalate ion in neutral aqueous solution, High Energ. Chem. 42 (2008) 1–6, https://doi.org/ 10.1134/s0018143908010013. [125] S.F. Lee, E. Jimenez-Relinque, I. Martinez, M. Castellote, Effects of Mott-Schottky frequency selection and other controlling factors on flat-band potential and bandedge position determination of TiO 2 , Catalysts 13 (2023) 1000, https://doi.org/ 10.3390/CATAL13061000. [126] N. Sakai, Y. Ebina, K. Takada, T. Sasaki, Electronic band structure of titania semiconductor nanosheets revealed by electrochemical and photoelectrochemical studies, J Am Chem Soc 126 (2004) 5851–5858, https://doi.org/10.1021/ JA0394582. [127] Y. Liu, L. Chen, X. Liu, T. Qian, M. Yao, W. Liu, H. Ji, Tuning band structure of graphitic carbon nitride for efficient degradation of sulfamethazine: Atmospheric condition and theoretical calculation, Chin. Chem. Lett. 33 (2022) 1385–1389, https://doi.org/10.1016/J.CCLET.2021.08.061. [128] Q. Su, Y. Li, R. Hu, F. Song, S. Liu, C. Guo, S. Zhu, W. Liu, J. Pan, Heterojunction photocatalysts based on 2D materials: the role of configuration, Adv Sustain Syst 4 (2020) 2000130, https://doi.org/10.1002/ADSU.202000130. [129] P.M. Wood, The potential diagram for oxygen at pH 7, Biochem. J 253 (1988) 287–289, https://doi.org/10.1042/BJ2530287. [130] K.S. Ranjith, D. Ranjith Kumar, S.M. Ghoreishian, Y.S. Huh, Y.K. Han, R. T. Rajendra Kumar, A radially controlled ZnS interlayer on ultra-long ZnO–Gd 2 S 3 core–shell nanorod arrays for promoting the visible photocatalytic degradation of antibiotics, Nanoscale 12 (2020) 14047–14060, https://doi.org/10.1039/ D0NR03094E. [131] W.H. Koppenol, D.M. Stanbury, P.L. Bounds, Electrode potentials of partially reduced oxygen species, from dioxygen to water, Free Radic Biol Med 49 (2010) 317–322, https://doi.org/10.1016/J.FREERADBIOMED.2010.04.011. R. Zandipak et al.