
==== Front
Proc Natl Acad Sci U S A
Proc Natl Acad Sci U S A
PNAS
Proceedings of the National Academy of Sciences of the United States of America
0027-8424
1091-6490
National Academy of Sciences

38478697
202318787
10.1073/pnas.2318787121
research-articleResearch Articleenv-sci-physEnvironmental Sciences417
Physical Sciences
Environmental Sciences
Harmonizing the cyano-group and Na to enhance selective photocatalytic O2 activation on carbon nitride for refractory pollutant degradation
Xu Mingkai a
Wang Ruizhao a b
Fu Haoyang a
Shi Yanbiao b
Ling Lan linglan@tongji.edu.cn
a 1 https://orcid.org/0000-0001-7348-4657

aState Key Laboratory for Pollution Control and Resource Reuse, College of Environmental Science and Engineering, Tongji University, Shanghai 200092, China
bSchool of Environmental Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
1To whom correspondence may be addressed. Email: linglan@tongji.edu.cn.
Edited by Alexis Bell, University of California, Berkeley, CA; received October 27, 2023; accepted January 3, 2024

13 3 2024
19 3 2024
13 9 2024
121 12 e231878712127 10 2023
03 1 2024
Copyright © 2024 the Author(s). Published by PNAS.
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).

Significance

Selectively activating O2 into free radicals with strong oxidation capacity through the charge-carrier transfer pathways is crucial for environmental remediation. Two-dimensional (2D) photocatalysts with short charge-carrier diffusion lengths have promising prospects for selective O2 activation. However, the free radical production of 2D photocatalysts is impeded by its strong exciton effect and insufficient charge-carrier mobility. This research presents a strategy of introducing the cyano-group and Na into graphitic carbon nitride to promote the charge-carrier transfer pathways. The cyano-group and Na simultaneously facilitate the exciton dissociation and strengthen the built-in electron field and electron sink effect of the catalyst, thus improving its selectivity of free radicals. This work provides a unique perspective on the design of catalysts used in selective O2 activation.

Manipulating exciton dissociation and charge-carrier transfer processes to selectively generate free radicals of more robust photocatalytic oxidation capacity for mineralizing refractory pollutants remains challenging. Herein, we propose a strategy by simultaneously introducing the cyano-group and Na into graphitic carbon nitride (CN) to obtain CN-Cy-Na, which makes the charge-carrier transfer pathways the dominant process and consequently achieves the selective generation of free radicals. Briefly, the cyano-group intensifies the local charge density of CN, offering a potential well to attract the hole of exciton, which accelerates the exciton dissociation. Meanwhile, the separated electron transfers efficiently under the robust built-in electric field induced by the cyano-group and Na, and eventually accumulates in the heptazine ring of CN for the following O2 reduction due to the reinforced electron sink effect caused by Na. As a result, CN-Cy-Na exhibits 4.42 mmol L−1 h−1 productivity with 97.6% selectivity for free radicals and achieves 82.1% total organic carbon removal efficiency in the tetracycline photodegradation within 6 h. Additionally, CN-Cy-Na also shows outstanding photodegradation efficiency of refractory pollutants, including antibiotics, pesticide plastic additives, and dyes. This work presents an innovative approach to manipulating the exciton effect and enhancing charge-carrier mobility within two-dimensional photocatalysts, opening an avenue for precise control of free radical generation.

selective O2 activation
free radicals
exciton effect
built-in electric field
electron sink effect
MOST | NSFC | National Outstanding Youth Foundation of China 501100010225 21822607 Lan Ling MOST | NSFC | Major Research Plan 501100010905 22176147 Lan Ling the Fundamental Research Funds for Central Universities 22120230295 Lan Ling
==== Body
pmcRapid industrialization has led to a concerning increase in environmental pollution, particularly refractory pollutants in water bodies (1–3). This alarming trend presents numerous challenges for both our environment and public health. Among various strategies to tackle this issue, photocatalytic activation of O2 into reactive oxygen species (ROS) has emerged as one of the most efficient approaches for refractory pollutant degradation (4). Generally, O2 can be converted into ROS through charge-carrier or exciton transfer pathways (Scheme 1) (5). The charge-carrier transfer pathways involve localized excitation of photoelectrons and holes, leading to the main product of free radicals [(superoxide radical (•O2–) and hydroxyl radical (•OH)]; in contrast, the exciton transfer pathways generate excitons in excited states that promote energy transfer to O2 and thus produce non-free radical [singlet oxygen (1O2)]. Compared to non-free radicals, the higher oxidation capacity of free radicals offers a superior strength for the effective mineralization of refractory pollutants (6–8). Therefore, the key to achieving high-efficiency photodegradation of refractory pollutants lies in finding ways to promote the charge-carrier transfer pathways for O2 activation.

Scheme 1. O2 activation pathways for semiconductor photocatalysts. (A) The charge-carrier transfer pathways involving charge-carrier separation and charge-carrier transfer from conduction band (CB) to MS. (B) The exciton transfer pathways involving photoexcitation of electron–hole pairs to form excitons and ISC from S1 to T1.

Typical O2 activation via charge-carrier transfer pathways primarily comprises photoexcitation, spatial electron–hole separation, and electron transfer to the adsorbed O2 molecules (Scheme 1A). Since photoexcited electrons tend to combine with the twinborn holes in the valence band, the spatial separation of the electron–hole pairs during activation is often considered the rate-determining step of the charge-carrier transfer pathways (9). One of the effective strategies to inhibit the recombination of electron−hole pairs in O2 activation is reducing the transportation distance of photogenerated charge carriers. Thus, the inherent advantages of two-dimensional (2D) semiconductors [e.g., CN (carbon nitride), MoS2, and BiOCl] with short charge-carrier diffusion lengths have gained significant attention (10–12). Unfortunately, the weakened dielectric screening in 2D photocatalysts results in stronger interlayer Coulomb interactions between photogenerated electrons and holes than the bulk photocatalysts, which impedes their O2 activation selectivity (13). Over the past decade, numerous approaches such as function group modification (14), surface doping (15), and vacancy engineering (16) have been devoted to manipulating the local charge density and relieving the exciton effect in 2D semiconductors. Although these strategies effectively alleviate the exciton effect of 2D materials, the separated electrons still tend to recombine with their twinborn holes after exciton dissociation, significantly hindering their O2 activation efficiency (17–19). Accordingly, suppressing the recombination of electron–hole pairs while weakening the exciton effect in 2D photocatalysts is highly desired for selective O2 activation.

In this study, we propose an approach by harmonizing the cyano-group and Na introduced into CN with a molten-salt method to simultaneously manipulate the exciton dissociation and electron–hole recombination of 2D photocatalysts. The local charge density exhibits dramatic enhancement after introducing the cyano-group, which offers a deep and narrow potential well to trap the hole of exciton and renders a space separation of the electron–hole pair. Thus, the electron of exciton shows extraordinary transfer mobility under enhanced built-in electron field induced by the cyano-group and Na. Thanks to the increased electron sink effect of Na, the transferred electron can be trapped by the heptazine rings of CN and then transferred to the adsorbed O2. Based on the above merits, the obtained CN-Cy-Na displays only 43 meV of exciton binding energy and thus exhibits superior selective generation of ROS, with the cumulative concentrations of •O2– and •OH in the CN-Cy-Na system 12.3 and 9.2 times higher than those of the pristine CN, respectively. The high concentrations of •O2– and •OH provide CN-Na-Cy with an extraordinary photocatalytic TC (tetracycline) mineralization efficiency (82.1% within 6 h), much higher than CN (29.2% within 6 h).

Results and Discussion

Synthesis and Structure Characterization.

The CN-Cy-Na photocatalyst was obtained by simultaneously introducing the cyano-group and Na into CN. The pristine CN was typically prepared by thermal polymerization using urea as the precursor. Subsequently, the pristine CN was treated by a molten-salt method (Fig. 1A). NaOH reacts with the amino group of CN at its melting point to generate the cyano-group and Na simultaneously by dehydration (SI Appendix, Fig. S1). The CN with the mono cyano-group (CN-Cy) or Na (CN-Na) were synthesized for comparison, and the detailed synthesis processes are provided in the SI Appendix, Experimental Procedures section.

Fig. 1. Characterization of CN-Cy-Na. (A) Schematic illustration of the preparation process for CN-Cy-Na. (B) TEM image, STEM image, and EDS elemental mappings of CN-Cy-Na. (C) C1s and N1s X-ray photoelectron spectroscopy (XPS) spectra, (D) Solid-state 13C MAS NMR spectra, and (E) FT-IR spectra of CN-Cy-Na, CN-Cy, CN-Na, and CN.

Transmission electron microscopy (TEM) images of CN-Cy-Na reveal a similar wrinkled microstructure of layered silk-like sheets compared to the pristine CN, which demonstrated that the nanosheet structure of CN is well reserved after molten-salt treatment (Fig. 1B and SI Appendix, Fig. S2). High-angle annular dark-field scanning TEM (HAADF-STEM) images and energy-dispersive spectroscopy (EDS) elemental mappings show that Na is uniformly dispersed on CN-Cy-Na and CN-Na. At the same time, no detectable Na signals are observed on CN-Cy and CN (Fig. 1B and SI Appendix, Fig. S3). These results suggest that the Na is successfully doped into CN on CN-Cy-Na, a finding further substantiated by the XPS result (SI Appendix, Fig. S4). The Na content of CN-Cy-Na and CN-Na is determined to be 2.8 wt.% and 2.2 wt.%, respectively, as indicated by inductively coupled plasma optical emission spectrometer (ICP-OES) analysis (SI Appendix, Table S1). The structural changes of CN after molten-salt treatment were then investigated by the XPS spectra of C1s and N1s. As shown in Fig. 1C, C 1s spectrum of all samples presents two peaks at 284.8 and 288.2 eV, assigned to adventitious hydrocarbons and N-C=N in the typical heterocyclic ring, respectively (20). CN and CN-Na exhibit higher intensity of N-C=N compared to CN-Cy-Na and CN-Cy, indicating the destruction of the heptazine framework after molten-salt treatment. A distinct peak at 286.5 eV attributed to the cyano-group is also observed on CN-Cy-Na and CN-Cy (21). N1s spectrum of CN and CN-Na exhibits three peaks attributed to bi-coordinated (N2C; 398.7 eV), tri-coordinated N (N3C; 400.2 eV), and -NHx groups (401.2 eV) in the heptazine framework, respectively (22). The N3C structure characteristic peaks of CN-Cy-Na and CN-Cy shift to lower binding energy (399.8 eV) compared to CN and CN-Na, possibly due to the structural changes induced by the cyano-group (23). Based on the XPS results, we could speculate that the -NH2 group at the apex of the heptazine structure on CN is dehydrogenized and transformed into a cyano-group following molten-salt treatment.

To further confirm the successful introduction of the cyano-group, the surface groups of the synthesized samples were characterized by solid-state 13C magic-angle spinning NMR (MAS NMR) and Fourier transform infrared (FT-IR) spectra (Fig. 1 D and E). The MAS NMR spectra of all samples show two strong peaks at 156 and 164 ppm, which are assigned to the CN3 (3) and CN2(-NHx) (2) groups in heptazine units (24). The peak (3) intensity of CN-Cy-Na and CN-Cy is significantly lower than that of CN and CN-Na, and two new peaks at 121 and 172 ppm attributed to the C atom in the cyano-group (1) and neighbor site (4) are observed in CN-Cy-Na and CN-Cy. (25) The FT-IR spectra of all samples show similar adsorption peaks at 810, 900 to 1,800, and 3,200 cm–1, corresponding to the bending vibration of heptazine rings, stretching vibration of N-C=N heterorings, and -NHx group, respectively (26). However, CN-Cy-Na and CN-Cy show weaker adsorption intensity of the -NHx group and exhibit a new peak at 2,177 cm–1, attributed to the asymmetric stretching vibration of the cyano-group (27). The above results further identify that the -NHx group at the apex of the melon structure is deprotonated into a cyano-group after molten-salt treatment. The chemical structure of synthetic samples was further characterized by X-ray diffraction. SI Appendix, Fig. S5 shows that two characteristic peaks at 13.1° and 27.3° corresponding to the (100) and (002) crystal planes of CN can be found for all samples (28). The (002) plane of CN-Cy-Na and CN-Cy displays reduced intensity and a subtle shift toward a lower angle, which concretely confirms that the heptazine rings are disrupted by molten-salt treatment and the generated cyano-group may result in an increased interlayer distance between the nanosheets. The enlargement of the interlayer distance is recognized for its ability to mitigate the spin-layer locking effect in 2D materials, thus facilitating the formation of interlayer excitons with reduced binding energy (29). In contrast, no significant shift is observed for the peak of the (002) plane on CN-Na, indicating that the interlayer distance of CN does not change after the introduction of Na. Therefore, Na is speculated to present in the framework rather than the interlayer of CN (SI Appendix, Fig. S6). Finally, nitrogen sorption analysis was used to evaluate the effect of NaOH molten salt on the specific surface area and pore structures of as-synthesized CN samples (SI Appendix, Fig. S7). CN and CN-Na exhibit higher Brunauer–Emmett–Teller specific surface area than CN-Cy and CN-Cy-Na. This result indicates a loss of mesoporous structure after molten-salt treatment, which further confirms that the amino group of the heptazine framework is transformed into the cyano-group and Na after molten-salt treatment.

Characterization of Electron–Hole Separation Efficiency.

To unveil the impact of the cyano-group and Na on the exciton dissociation, the local charge density of different samples was first examined by electrostatic potential (ESP) calculation. As illustrated in Fig. 2A and SI Appendix, Fig. S8, the region near the cyano-group processes higher charge density than the rest of the area. These results demonstrate that introducing the cyano-group can significantly enhance the local charge density of CN, which is essential for exciton dissociation (30). This finding is further supported by Kelvin probe force microscopy (KPFM) results. As shown in Fig. 2B, CN-Cy-Na and CN-Cy exhibit superior surficial ESP of 35.4 and 23.1 mV, respectively, 2.2 and 1.5 folds to that of CN (15.9 mV). On the other hand, CN-Na shows a slight enhancement in surficial ESP (18.8 mV). These results further suggest that the cyano-group can significantly enhance the local charge density of CN. Subsequently, we recorded the steady-state photoluminescence (PL) spectra at different temperatures (from 15 to 300 K) to calculate the Eb of different samples with the Arrhenius equation:IT=I01+A exp(-EbkBT),

Fig. 2. Characterizations of exciton dissociation efficiencies. (A) The ESP calculation result of CN-Cy-Na and CN-Na. (B) The KPFM images of CN-Cy-Na, CN-Cy, CN-Na, and CN. (C) Steady-state PL spectra with a function of temperature of as-prepared CN photocatalysts (λexcitation = 375 nm).

where I0 stands for the PL intensity at 0 K; T represents the temperature; and kb refers to the Boltzmann constant. The results presented in Fig. 2C show that CN exhibits the highest Eb of 78 meV, consistent with the previous work (31). With the introduction of the cyano-group, Eb significantly decreases to 43 meV for CN-Cy-Na and 57 meV for CN-Cy. However, CN-Na shows a slight decrease in Eb, measured at 69 meV, compared to the pristine CN. This indicates that the enhanced local charge density induced by cyano-groups offers a deep and narrow potential well to trap the holes of exciton, hence weakening the exciton effect of CN.

As aforementioned, although reducing Eb enables the rapid dissociation of exciton into the charge carriers and facilitates the charge-carrier transfer pathways, the O2 activation efficiency is still impeded by the recombination of electron–hole pairs, which is strongly relative to the charge-carrier mobility and electron sink effect of semiconductor (14, 32). Generally, higher charge-carrier mobility offers spatial separation of charge carriers and consequently hinders the recombination of electron–hole pairs; meanwhile, a stronger electron sink effect renders the opportunity to restore the electron for subsequent reaction, which prolongs the lifetime of charge carriers and facilitates the electron transfer from photocatalysts to O2. Accordingly, we estimated the charge-carrier mobility of different CN samples by measuring the electrochemical impedance spectroscopy and transient photocurrent of all synthetic samples. It is noted that CN-Cy-Na shows smaller arc radii of the Nyquist plot than its counterparts, suggesting that it has the lowest charge transfer resistance and highest charge-carrier mobility (SI Appendix, Fig. S9). Additionally, CN-Cy-Na shows significantly higher photocurrent density than all its counterparts, indicating that introducing the cyano-group and Na can synergistically improve the transfer efficiency of charge carriers (Fig. 3A). Based on the above analysis, the built-in electric field intensity of different samples was calculated through a Kanada model:E=-2Vsρεε0,

Fig. 3. Characterizations of charge-carrier mobility and electron sink effect. (A) Transient photocurrent responses, (B) built-in electric field intensity, and (C) time-resolved PL spectra of different photocatalysts. (D) Schematic illustration of the electron sink model and the calculated plane-averaged potential of different photocatalysts. COHP of (E) CN-Cy-Na and (F) CN-Cy. (G) O2-TPD signals of as-prepared photocatalysts.

where Vs represents the surface potential measured by KPFM; ρ refers to the surface charge density obtained through current density (SI Appendix, Note S1); ε stands for the dielectric constant detected by the vector network analyzer (SI Appendix, Fig. S10); and ε0 is the vacuum dielectric constant. As shown in Fig. 3B, the intensity of the built-in electric field in CN-Cy-Na is calculated to be 5.1, 4.1, and 2.2 times that of CN, CN-Na, and CN-Cy, respectively. Such significantly higher intensity in CN-Cy-Na is further confirmed by the room-temperature steady-state PL spectrum (SI Appendix, Fig. S11). Additionally, CN-Cy-Na displays a much longer lifetime (τ) of 4.09 ns, which is nearly four times that of CN (Fig. 3C). These results provide concrete evidence that introducing the cyano-group and Na into CN markedly enhances the intensity of the built-in electric field, resulting in improved carrier mobility and a longer lifetime for charge carriers.

Subsequently, the electron sink effect of synthesized samples was measured by multiple methods, including theoretical calculations and O2 temperature-programmed desorption (O2-TPD) experiments to estimate their electron-accepting/donating ability. The work functions were first measured by the pane-averaged potential to estimate the electron-accepting abilities of different samples. As shown in Fig. 3D, CN-Cy-Na and CN-Na possess deeper and narrower potential well compared to their Na-free counterparts. This result suggests that introducing Na can improve the electron-accepting ability of heptazine rings in CN to trap the separated electron, which hinders the recombination of electron–hole pairs and facilitates the subsequent O2 reduction (33). Then, we calculated the electron transfer process between photocatalyst and O2 to investigate the impact of Na on the electron-donating ability of CN. The density of states shows that CN-Cy-Na and CN-Na ion exhibit higher Femi-level than CN-Cy and CN, indicating more effortless electron transfer from photocatalyst to adsorbed O2 (SI Appendix, Fig. S12). As expected, the charge density difference of CN exhibits neglected electron transfer between the adsorbed O2 and photocatalyst, while O2 is more likely to extract electrons from the photocatalyst with Na (SI Appendix, Fig. S13). The electron distribution of O2 orbitals adsorbed on different samples was calculated by the crystal orbital Hamilton populations (COHP) analyst. As shown in Fig. 3 E and F and SI Appendix, Fig. S14, the CN-Cy-Na and CN-Na possess more electrons in the π antibonding orbital of O2. The bonding strength of O-O bond was further estimated by the integrated COHP, which illustrates that the O-O bond strength is significantly weaker for O2 adsorbed on the Na site. This result suggests that CN with Na exhibits stronger electron-donating ability and easier population for the π antibonding orbital of O2. Accordingly, O2 displays higher Eads after introducing Na (SI Appendix, Fig. S15). These theoretical results are further qualitatively confirmed by O2-TPD. As shown in Fig. 3G, the desorption curve of CN exhibits a maximum at 116 °C ascribed to the desorption of O2, and the desorption peak shifts to a higher temperature after the introduction of Na. These results demonstrate that the electron sink effect of CN is dramatically enhanced with the introduction of Na.

O2 Activation Test.

Introducing the cyano-group and Na significantly enhances the local charge density and electron sink effect of CN, CN-Cy-Na is expected to exhibit better O2 activation efficiency through the charge-carrier transfer pathways. Accordingly, we first evaluated the ROS generated under light irradiation using electron spin resonance spectroscopy (ESR). All samples were stirred in the dark for 30 min, and no ROS was detected in any of the photoreaction systems (SI Appendix, Fig. S16). Once the light was turned on, three different species (•O2–, 1O2, and •OH) were generated by all photocatalysts (Fig. 4 A–C). As expected, CN-Cy-Na exhibits the highest intensity of •O2– and •OH signals among all catalysts. The intensity of 1O2 decreases in the order of CN, CN-Na, CN-Cy, and CN-Cy-Na, opposite the intensity ordering of •O2– and •OH. The concentrations of different ROS were further qualified by using nitro blue tetrazolium (NBT), salicylic acid, and 1,3-diphenylisobenzofuran (DPBF) as the probes of •O2–, •OH, and 1O2 (16, 34, 35). As shown in Fig. 4 D–F, CN-Cy-Na displays an impressive free radicals productivity of 4.42 mmol L–1 h–1, which is 11.84 folds that of pristine CN, while CN possesses much higher 1O2 productivity than other photocatalysts. The selectivity of each ROS was further calculated, and the results are displayed in Fig. 4G. Strikingly, CN-Cy-Na possesses 97.6% selectivity of free radicals, which is significantly higher than CN (58.6%), indicating that CN-Cy-Na is inclined to activate O2 through the charge-carrier transfer pathways. The action spectra of CN-Cy-Na and CN for free radical formation were further measured under monochromatic light irradiation, where CN-Cy-Na displays significantly higher apparent quantum yields (ΦAQY) than CN at all of the wavelengths (SI Appendix, Fig. S17). Finally, the origin of ROS was tested by measuring the production of different ROS within 60 min under various gas feeding environment (SI Appendix, Fig. S18). The concentration of ROS is significantly improved by O2 and negligible ROS is detected under N2 feeding, which indicates that O2 is the main oxygen source of ROS.

Fig. 4. Photocatalytic O2 activation and ROS qualification. The ESR spectra of (A) DMPO-•O2–, (B) DMPO-•OH, and (C) TEMP-1O2 adducts for different photocatalysts under light irradiation. Quantitative determination for the ROS amount of (D) •O2–, (E) •OH, and (F) 1O2. (G) The calculated selectivity of free radicals for different photocatalysts. (H) Schematic illustration of charge-carrier transfer pathways on CN-Cy-Na.

Based on the above results, we conclude that CN-Cy-Na can achieve selective O2 activation with high efficiency for the following reasons (Fig. 4H): 1) The cyano-group can increase the local charge density of CN to extract the holes in exciton and thus weaken the exciton effect; 2) The separated charge carriers can be transferred efficiently under the robust built-in electric field brought by the cyano-group and Na; 3) The transferred electron can be reserved by the heptazine rings of CN through the electron sink effect enhanced by Na, which retards the electron–hole pair recombination and promotes the following O2 reduction; 4) The Na sites can significantly improve the electron transfer from photocatalysts to the adsorbed O2, thus facilitating reduction of O2 into free radicals.

Photocatalytic Degradation of TC.

After that, TC was selected as a model pollutant to evaluate the photocatalysis behavior for refractory pollutants degradation using a PerfectLight PCX-50C Discover multichannel photocatalytic reaction system (SI Appendix, Fig. S19). As shown in SI Appendix, Fig. S20, all samples show limited efficiency in removing TC without light. Batch photodegradation experiments were performed after the adsorption-desorption equilibrium was reached. As shown in Fig. 5A, CN-Cy-Na completely removes TC within 60 min, while CN, CN-Na, and CN-Cy only achieve 57.3%, 78.1%, and 83.1% degradation efficiency, respectively. All photodegradation curves follow the pseudo-first-order reaction kinetic model. The kinetic constant of CN-Cy-Na is calculated to be 6.8 × 10–2 min–1, which is much higher than those of CN (1.4 × 10–2 min–1), CN-Na (2.6 × 10–2 min–1), and CN-Cy (2.9 × 10–2 min–1) (Fig. 5B). This result suggests that CN-Cy-Na has better photodegradation ability toward TC than all counterparts. To further elucidate the effect of ROS species on reaction activity, quenching experiments were also performed in the reaction system of CN-Cy-Na and CN (Fig. 5C). Superoxide dismutase (SOD), tert-Butyl alcohol, catalase (CAT), and β-carotene were selected as the scavenger of •O2–, •OH, H2O2, and 1O2, respectively. For CN, SOD and β-carotene show similar inhibit efficiency, while tert-Butyl alcohol and CAT only show a slight effect on TC photodegradation, which suggests that •O2– and 1O2 are the dominated ROS in the CN system. However, SOD suppresses most TC photodegradation in the CN-Cy-Na system, implying that •O2– is the primary ROS account for TC degradation. This result is consistent with the ESR experiments, which emphasizes the importance of ROS selectivity on refractory pollutant photodegradation. Then, the mineralization efficiency of different photocatalysts was evaluated by the total organic carbon (TOC) measurement. As shown in Fig. 4D, CN-Cy-Na removes 82.1% of TOC after 6 h of photodegradation, while only 29.2%, 52.5%, and 59.8% of TC are mineralized by CN, CN-Na, and CN-Cy, respectively after 6 h. The TOC removal efficiency of CN-Cy-Na outperforms that of most other reported photocatalysts (SI Appendix, Table S2). This result further confirms the outstanding photocatalytic activity of CN-Cy-Na and underscores the significance of the synergistic effect between the cyano-group and Na.

Fig. 5. Photocatalytic performance of as-prepared catalysts. (A) Photodegradation curves and (B) the corresponding kinetic plots for TC degradation with different catalysts. (C) Photocatalytic degradation efficiency of TC by CN and CN-Cy-Na with different scavengers. (D) TOC removal efficiency for TC degradation by different photocatalysts. (E) Photocatalytic degradation efficiency of various refractory pollutants using as-prepared photocatalysts.

To investigate the reusability and stability of CN-Cy-Na, the cyclic photodegradation activity of CN-Cy-Na for TC was evaluated by an 8-cycle experiment (SI Appendix, Fig. S21). After eight replicated tests, CN-Cy-Na shows a negligible decrease in photocatalytic activity for TC degradation. The FT-IR and HAADF-STEM results of CN-Cy-Na after the 8-cycle experiment are almost identical to those of the as-synthesized sample (SI Appendix, Figs. S22 and S23). Moreover, the leaching of Na from CN-Cy-Na was also monitored by ICP-OES analysis after each cycle, and only a trace amount of Na was detected (SI Appendix, Fig. S24). The above results suggest that CN-Cy-Na exhibits extraordinary stability and reusability under reaction conditions. The photocatalytic performance of CN-Cy-Na for TC degradation was further measured under a wide range of pH values (3–10). As shown in SI Appendix, Fig. S25, CN-Cy-Na keeps its high TC removal efficiency when pH ≥ 5. A slight decrease in photocatalytic activity at pH = 3 might be related to the leaching of Na. This result suggests that Na plays a critical role in the photodegradation process, further confirming the stability of CN-Cy-Na under different reaction conditions. Besides TC, CN-Cy-Na exhibits superior photocatalytic activity to all its counterparts in the degradation of other refractory pollutants, including bisphenol A, norfloxacin, methylene blue, and 4-chlorophenol (Fig. 5E). Other alkali/alkaline hydroxides are also applicable for the molten-salt treatment of CN, indicating the universal of our strategy (SI Appendix, Fig. S26). Thereafter, the photocatalytic degradation pathways of TC were investigated by liquid chromatography–mass spectrometry (SI Appendix, Fig. S27). Based on the product analysis, we proposed three possible pathways for the photocatalytic degradation of TC (SI Appendix, Fig. S28). For pathway I, the TC molecule is initially attacked by •OH, leading to demethylation and the formation of TP1 with an m/z of 416. TP1. Subsequently undergoes deamidation to yield TP2 with an m/z of 358, dehydrated and demethylated to produce TP3 (m/z = 326) and TP4 (m/z = 312). In pathway II, the N-dealkyl and acylamino groups of TC are first attacked by •OH to produce TP5 (m/z = 378). With further attack by •O2–, TP5 opens the loop to produce TP6 (m/z = 382) and TP7 (m/z = 232). For pathway III, TC was directly oxidized by •O2– to produce TP8 (m/z = 371) through the ring-opening and deamidation processes. Afterward, the ring-opening product TP8 undergoes decarboxylation, dealkylation, and deamination to generate TP9 (m/z = 288), TP10 (m/z = 246), and TP11 (m/z = 230). Finally, the resulting TP4, TP7, and TP11are further mineralized into a series of small molecules [P12 (m/z = 110), P13 (m/z = 132), P14 (m/z = 56), P15 (m/z = 113], and even CO2 and H2O. Comparative biological toxicity study on the degradation products of CN-Cy-Na and CN was performed using Photobacterium phosphoreum T3 as the acute toxicity indicator and the results show that the acute toxicity of the degradation products of CN-Cy-Na is significantly lower than that of CN (SI Appendix, Fig. S29).

In conclusion, we demonstrate that a selective O2 activation can be achieved by harmonizing the cyano-group and Na introduced into CN to simultaneously manipulate the exciton dissociation and charge-carrier transfer processes of photocatalysis. The addition of the cyano-group immensely increases the local charge density and facilitates the exciton dissociation process in CN. The separated charge-carriers transfer efficiently under a robust built-in electric field. These charge carriers are finally trapped by the heptazine rings of CN, as Na enhances the electron sink effect of CN, which suppresses the recombination of electron–hole pairs and improves the electron transfer from photocatalyst to O2. Accordingly, CN-Cy-Na displays 97.6% free radical selectivity, significantly higher than CN (58.6%). As expected, CN-Cy-Na also exhibits superior photocatalytic activity to all its counterparts on the degradation of refractory pollutants, including antibiotic, pesticide, plastic additive, and dye. This study proposes a facile strategy to manipulate the electron–hole pair separation process within 2D photocatalysts for excellent pollutant control with solar energy in the controllable selective pathway.

Materials and Methods

Preparation of Photocatalysts.

The pristine CN was prepared by the pyrolysis treatment of urea. Typically, 10 g of urea powder was placed in a crucible and calcined in a muffle furnace at 550 °C for 4 h with a heating rate of 10 °C/min. After cooling down to room temperature, the yellow powder was collected and ground for further use. Simultaneously grafting of the cyano-group and Na on CN was achieved by the molten-salt treatment. Briefly, 5 g of CN and 1 g NaOH were dissolved with stirring into 30 mL H2O. The resulting solution was sonicated for 30 min and vacuum dried at 50 °C. Then the mixture was thoroughly ground for 20 min under an infrared lamp and calcined in a muffle furnace for 2 h with a heating rate of 10 °C/min. The as-obtained reddish sample was washed with DI water, ethanol and dried. The final product was denoted as CN-Na-Cy. For comparison, cyano-group-only and Na-only modified CN was also synthesized. Briefly, the cyano-group-only modified CN was prepared by washing CN-Cy-Na with dilute H2SO4 (0.01 mol/L) and DI water to remove Na, the as-obtained sample was denoted as CN-Cy. Na-only modified CN was obtained by treating 0.2 g of CN in 100 mL of 0.3 mol/L NaOH solution at 60 °C for 4 h, the product was washed with DI water, ethanol and dried, the final product was denoted as CN-Na.

ROS Measurement.

The amount of •OH was determined by the salicylic acid method. Specifically, salicylic acid could react with hydroxyl radicals to produce phenol (11%), 2,3-dihydroxybenzoic (49%), and 2,5-dihydroxybenzoic acid (40%). The concentration of 2,5-dihydroxybenzoic acid was monitored by an ultraviolet spectrophotometer at λmax = 330 nm. The amount of •O2– was quantified by monitoring the concentration of NBT as NBT could reduce by •O2– to form formazan. The concentration of NBT was measured by detecting the absorbance at λmax = 259 nm. 1,3-DPBF was used as the probe of 1O2. DPBF could react to 1O2 with a molar ratio of 1:1, and the concentration of DPBF was tested at 410 nm using an ultraviolet spectrophotometer. ESR spectra of different radicals were performed on the Bruker EMXplus-6/1 spectrometer using aqueous and methanolic as media. 5-dimethyl-1-pyrroline-N-oxide (DMPO) was used as the spin trapping agent for •OH and •O2–, while 2,2,6,6-tetramethylpiperidine (TEMP) was used as 1O2 trapper. Quenching experiments were performed by introducing a series of specific scavengers into the reaction system. Typically, SOD (200 units/mL), isopropanol (2 mmol/L), CAT (200 units/mL), β-carotene (0.2 mmol/L), and EDTA-2Na (1 mmol/L) were used to quench •O2–, •OH, H2O2, 1O2, and h+, respectively.

Theoretical Calculations.

Detailed information is shown in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

This work was supported by the National Outstanding Youth Science Fund Project of the National Natural Science Foundation of China (No. 21822607), the National Natural Science Foundation of China (No. 22176147), and the Fundamental Research Funds for Central Universities (No. 22120230295).

Author contributions

L.L. supervised the research; M.X., R.W., and L.L. planned the research; M.X. and R.W. performed research; M.X. and Y.S. analyzed the data; M.X., H.F., and L.L. wrote the article; and all the authors participated in the interpretation and discussion of the results.

Competing interests

The authors declare no competing interest.

Data, Materials, and Software Availability

All the study data are included in the article and/or SI Appendix.

Supporting Information

This article is a PNAS Direct Submission.
==== Refs
1 M. Ran , Selective production of CO from organic pollutants by coupling piezocatalysis and advanced oxidation processes. Angew. Chem. Int. Ed. 62 , e202303728 (2023).
2 Z. Chen , Single-atom Mo–Co catalyst with low biotoxicity for sustainable degradation of high-ionization-potential organic pollutants. Proc. Natl. Acad. U.S.A. 120 , e2305933120 (2023).
3 J. Xu , Sulfur dose and sulfidation time affect reactivity and selectivity of post-sulfidized nanoscale zerovalent iron. Environ. Sci. Technol. 53 , 13344–13352 (2019).31622083
4 H. Zhan, R. Zhou, P. Wang, Q. Zhou, Selective hydroxyl generation for efficient pollutant degradation by electronic structure modulation at Fe sites. Proc. Natl. Acad. U.S.A. 120 , e2305378120 (2023).
5 D. Zhang , Tailoring of electronic and surface structures boosts exciton-triggering photocatalysis for singlet oxygen generation. Proc. Natl. Acad. U.S.A. 118 , e2114729118 (2021).
6 X. Duan, H. Sun, Z. Shao, S. Wang, Nonradical reactions in environmental remediation processes: Uncertainty and challenges. Appl. Catal. B Environ. 224 , 973–982 (2018).
7 Q. Zhou , Generating dual-active species by triple-atom sites through peroxymonosulfate activation for treating micropollutants in complex water. Proc. Natl. Acad. U.S.A. 120 , e2300085120 (2023).
8 J. Qu , Free radicals-triggered reductive and oxidative degradation of highly chlorinated compounds via regulation of heat-activated persulfate by low-molecular-weight organic acids. Appl. Catal. B Environ. 310 , 121359 (2022).
9 Y. Shi , Surface boronizing can weaken the excitonic effects of BiOBr nanosheets for efficient O2 activation and selective NO oxidation under visible light irradiation. Environ. Sci. Technol. 56 , 14478–14486 (2022).36173086
10 T. Jia , Single-atom nickel on carbon nitride photocatalyst achieves semihydrogenation of alkynes with water protons via monovalent nickel. Angew. Chem. Int. Ed. 62 , e202216511 (2023).
11 X. Li, J. Chen, X. Wang, J. Sun, M. Antonietti, Metal-free activation of dioxygen by graphene/g-C3N4 nanocomposites: Functional dyads for selective oxidation of saturated hydrocarbons. J. Am. Chem. Soc. 133 , 8074–8077 (2011).21561075
12 H. Xu, X. Liu, H. Li, L. Zhang, O2 activation and 1O2 generation over phosphate modified BiOCl for efficient photodegradation of organic pollutants. Appl. Catal. B Environ. 314 , 121520 (2022).
13 Y. Shi , Manipulating excitonic effects in layered bismuth oxyhalides for photocatalysis. ACS EST Engg. 2 , 957–974 (2022).
14 J. Xu , Efficient photocatalytic hydrogen and oxygen evolution by side-group engineered benzodiimidazole oligomers with strong built-in electric fields and short-range crystallinity. Angew. Chem. Int. Ed. 61 , e202212243 (2022).
15 K. P. Musselman , Improved exciton dissociation at semiconducting polymer: ZnO donor: Acceptor interfaces via nitrogen doping of ZnO. Adv. Funct. Mater. 24 , 3562–3570 (2014).25520604
16 H. Wang , Boosting hot-electron generation: Exciton dissociation at the order–disorder interfaces in polymeric photocatalysts. J. Am. Chem. Soc. 139 , 2468–2473 (2017).28102077
17 L. B. Hoch , Carrier dynamics and the role of surface defects: Designing a photocatalyst for gas-phase CO2 reduction. Proc. Natl. Acad. U.S.A. 113 , E8011–E8020 (2016).
18 Y. Qian, D. Li, Y. Han, H. Jiang, Photocatalytic molecular oxygen activation by regulating excitonic effects in covalent organic frameworks. J. Am. Chem. Soc. 142 , 20763–20771 (2020).33226795
19 L. Yu , Locally asymmetric BiOBr for efficient exciton dissociation and selective O2 activation toward oxidative coupling of amines. ACS Nano 17 , 15077–15084 (2023).37489696
20 J. Fu , Hierarchical porous O-doped g-C3N4 with enhanced photocatalytic CO2 reduction activity. Small 13 , 1603938 (2017).
21 W. Wang , Potassium-ion-assisted regeneration of active cyano groups in carbon nitride nanoribbons: Visible-light-driven photocatalytic nitrogen reduction. Angew. Chem. Int. Ed. 58 , 16644–16650 (2019).
22 Z. Lan, G. Zhang, X. Wang, A facile synthesis of Br-modified g-C3N4 semiconductors for photoredox water splitting. Appl. Catal. B Environ. 192 , 116–125 (2016).
23 H. Yu , Alkali-assisted synthesis of nitrogen deficient graphitic carbon nitride with tunable band structures for efficient visible-light-driven hydrogen evolution. Adv. Mater. 29 , 1605148 (2017).
24 B. JuÌˆrgens, , An important intermediate during condensation of melamine rings to graphitic carbon nitride: Synthesis, structure determination by X-ray powder diffractometry, solid-state NMR, and theoretical studies. J. Am. Chem. Soc 125 , 10288–10300 (2003).12926953
25 V. W. Lau , Rational design of carbon nitride photocatalysts by identification of cyanamide defects as catalytically relevant sites. Nat. Commun. 7 , 12165 (2016).27387536
26 X. Li, X. Wang, M. Antonietti, Solvent-free and metal-free oxidation of toluene using O2 and g-C3N4 with nanopores: Nanostructure boosts the catalytic selectivity. ACS Catal. 2 , 2082–2086 (2012).
27 W. Lei, D. Portehault, R. Dimova, M. Antonietti, Boron carbon nitride nanostructures from salt melts: Tunable water-soluble phosphors. J. Am. Chem. Soc 133 , 7121–7127 (2011).21506566
28 A. Thomas , Graphitic carbon nitride materials: Variation of structure and morphology and their use as metal-free catalysts. J. Mater. Chem. 18 , 4893–4908 (2008).
29 S. Zheng , Coupling and interlayer exciton in twist-stacked WS2 bilayers. Adv. Opt. Mater. 3 , 1600–1605 (2015).
30 Q. Hu , Observation of eydberg moiré excitons. Science 380 , 1367–1372 (2023).37384701
31 Y. Zhang , Metal-free N-GQDs/P-g-C3N4 photocatalyst with broad-spectrum response: Enhanced exciton dissociation and charge migration for promoting H2 evolution and tetracycline degradation. Sep. Purif. Technol. 304 , 122297 (2023).
32 Z. Zhang, S. Cao, Y. Liao, C. Xue, Selective photocatalytic decomposition of formic acid over AuPd nanoparticle-decorated TiO2 nanofibers toward high-yield hydrogen production. Appl. Catal. B Environ. 162 , 204–209 (2015).
33 Q. Wu , A metal-free photocatalyst for highly efficient hydrogen peroxide photoproduction in real seawater. Nat. Commun. 12 , 483 (2021).33473132
34 L. Ye, L. Zan, L. Tian, T. Peng, J. Zhang, The 001 facets-dependent high photoactivity of BiOCl nanosheets. Chem. Commun. 47 , 6951–6953 (2011).
35 J. Jiang , Photo-Fenton degradation of emerging pollutants over Fe-POM nanoparticle/porous and ultrathin g-C3N4 nanosheet with rich nitrogen defect: Degradation mechanism, pathways, and products toxicity assessment. Appl. Catal. B Environ. 278 , 119349 (2020).
