Unraveling the mechanisms of S-doped carbon nitride for photocatalytic oxygen reduction to H2O2

Yawen Tong a, Changgeng Wei a, Yi Li ab, Yongfan Zhang ab and Wei Lin *ab
aState Key Laboratory of Photocatalysis on Energy and Environment, College of Chemistry, Fuzhou University, Fuzhou 350108, P. R. China. E-mail: wlin@fzu.edu.cn
bFujian Provincial Key Laboratory of Theoretical and Computational Chemistry, Xiamen, Fujian 361005, China

Received 2nd July 2020 , Accepted 4th September 2020

First published on 5th September 2020

An in-depth understanding of the microscopic reaction mechanism on a nonmetal-doped catalytic system at the atomic level is one of the critical approaches to developing new efficient catalysts. Herein, the effects of S-doping on melon-based carbon nitride (CN) for the photocatalytic selective oxygen reduction reaction (ORR) have been comprehensively investigated by first-principles calculations. The configurations, electronic properties, optical properties, and the reaction performance of the S-doped melon-based CN have been studied and discussed. The results demonstrate that the decoration with S atoms exhibited substantial effects, involving the redistribution of the charge density and tuning of the bandgap, which promote the photocatalytic selective ORR activity. Accordingly, O2 is activated on the S-doped system with about 0.4 e of charge obtained from catalytic surfaces, leading to the thermodynamically feasible H2O2 and H2O formation, which is in good agreement with the experimental results. Our results provide theoretical insights into the design and development of polymeric carbon nitride (PCN) as well as other metal-free photocatalysts for the selective ORR.

1. Introduction

Hydrogen peroxide (H2O2) is environment-friendly peroxide because it generates harmless H2O only as a by-product during the reaction process. It is also a multi-functional peroxide applied in various fields, such as textiles and pulp bleaching, wastewater treatment, chemical synthesis, and environmental protection.1 H2O2 production can be generated by the anthraquinone auto-oxidation process,2 or the electrocatalytic reduction of O2via the proton-coupled electron transfer (PCET) method (O2 + 2H+ + 2e → H2O2; U° = 0.70 VRHE).3–5 Currently, the photocatalytic production of H2O2 through PCET by semiconductors has been an intensive endeavor worldwide, due to conditions of the photocatalytic reaction that are relatively mild with low-cost input of directly capturing, converting and storing sustainable solar energy.6–12

Most metal oxides, especially TiO2, have been widely studied for H2O2 photosynthesis.13–17 However, because the bandgap of anatase TiO2 is relatively wide (3.2 eV), it only responds to the ultraviolet range of the spectrum, indicating that only 4% of the incoming solar irradiation can be maximally harnessed. Therefore, it is highly desirable to explore novel visible-light-driven materials that have better photocatalytic performance or take innovative measures to modify known potential photocatalysts for improving their photoreactivity.

Since the pioneering work reported on H2 production that it can be generated by photocatalytic H2O splitting on PCN in 2009,18 PCN, composed of earth-abundant C and N elements has attracted extensive attention and is regarded as a promising potential photocatalyst mainly due to its distinct advantages of being environment friendly, excellent stability, fascinating electronic structure with suitable bandgap and position, as well as easy synthesis and modification.19–21 Nevertheless, two intrinsic flaws of PCN cannot be ignored, namely, the comparatively wide bandgap and the fast recombination of photogenerated charge carriers, which are crucial factors limiting photocatalytic reactions.20 Hence, significant endeavors have been made to enhance the photocatalytic activity of PCN by structural modification strategies, including morphology control, construction of a composite, loading co-catalysts, introduction of doping, etc.22–27 The non-metal doping of PCN maintains its original metal-free properties with cost-effective and environmental benignity,28 so great and fruitful efforts have been devoted to investigating its photocatalytic capacity after modification by non-metal doping, such as B, C, O, P, S doping, etc.29–35 The experimental studies have demonstrated that the modification of PCN with S-doping is an effective approach to improving the optical properties and photocatalytic behaviors.35–39 For instance, Liu et al.35 found that the H2 yield for S-doped PCN was 7.2–8 times higher than that of PCN. Additionally, Wang et al.36 reported that the yields of CH3OH produced by photocatalytic CO2 reduction on S-doped PCN (1.12 μmol g−1) were better as compared to the undoped CN (0.81 μmol g−1), and density functional theory (DFT) calculations37 confirmed this conclusion and explained the specific mechanism of the CO2 reduction reaction. Chuaicham et al.38 recently determined that S-doped PCN showed greater photocatalytic efficiency than PCN for the Cr(VI) reduction reaction. Furthermore, our recent collaborative work has demonstrated that under visible light irradiation, the photochemical production of H2O2 on S-doped melon-based CN (a kind of PCN that has been mostly studied) with the addition of controllable amounts of KCl and NaOH was achieved at a millimolar level per hour, along with almost 100% apparent quantum yield.39

To unravel how S-doped melon-based CN has a much better H2O2 yield on the molecular level, and why it delivers a superior photoreduction capacity as compared to melon-based CN, the effects of the altered component by the decoration of S atoms within the melon-based CN on its photocatalytic performance have been investigated at the atomic scale by exploring the mechanism of the (selective) ORR on S-doped samples. First, the configurations, electronic and optical properties of S-doped systems were comprehensively studied and discussed. The activation of O2 in all kinds of systems was also analyzed. We also calculated the Gibbs free energy change (ΔG) of the (selective) ORR paths to understand the microscopic mechanism. This study provides effective insight into the design and development of PCN, as well as other metal-free photocatalysts for the ORR.

2. Computation methods

The periodic spin-polarized DFT calculations on S-doped melon-based CN were carried out by using the VASP code40,41 with the Perdew–Burke–Ernzerhof (PBE) functional42 of the generalized gradient approximation (GGA). The cutoff energy was set to 520 eV for the plane wave expansion of wavefunctions. The DFT-D2 method developed by Grimme was used to consider van der Waals interactions.43 All configurations were fully optimized until the energy difference and force were less than 10−4 eV and 0.01 eV Å−1, respectively. The unit cell (1 × 1 × 1) was repeated periodically in the xy plane, while a vacuum region of 10 Å was set along the z-direction to avoid interactions with adjacent layers. The K-point of the Brillouin zone was sampled by using a (5 × 5 × 1) grid.44 The calculations of ΔG on each elementary reaction adopted the computational hydrogen electrode model (CHE) proposed by Nørskov et al.45–47 The formula for ΔG45,48 is defined as
where ΔE is the change in the electronic energy calculated by DFT, ΔEZPE is the change in the zero-point energy, T is the room temperature (298.15 K) and ΔS is the entropy change. ΔGpH is the free energy associated with pH, which is defined as ΔGpH = kBT × ln[thin space (1/6-em)]10 × pH, where kB is the Boltzmann constant, and pH = 0 is considered in this work.
ΔGU = −neU,
where n and U are the number of electrons transferred and the electrode potential applied, and the value of U was set to be zero. The free energy of the electron–hole pair was considered equal to that of 1/2H2, regarding the reaction H+ + e → 1/2H2 at standard conditions (p = 1 bar, T = 298.15 K, U = 0, pH = 0). The free energy of H2O was calculated at 0.035 bar pressure to imitate liquid water. The energy of O2 cannot be precisely calculated by the DFT approach as the result possesses complex electronic properties. Hence, the free energy of O2 was determined by the free energy change of the reaction 2H2 + O2 → 2H2O, which was 4.92 eV.49 The zero-point energy of the intermediates during the ORR was obtained by calculating the vibration frequency (Table S1, ESI). The calculation model adopts the method that the adsorbed species is released and the substrate is fixed, provided that the substrate vibration is insignificant. The entropy of the gas phase molecules was obtained from the NIST database. The adsorption energies (Eads)50,51 of the adsorbed species were described by the following equation:
Eads = E(total) − E(adsorbate) − E(surface)
where E(total), E(adsorbate), E(surface) are the total energies of the adsorbate–surface complex, the adsorbate and surface, respectively.

3. Results and discussion

3.1. Geometry structures and thermodynamic stability

In principle, C or N can be replaced by S atoms on melon-based CN. In our previous work,39 it was discussed that the doped system formed by replacing C with S requires much more energy than replacing N with S, which is consistent with the Experimental results where the S-doped system was preferably formed by the replacement of N with S.35 Our previous work showed that the system of C substituted by S cannot activate O2 at all.39 Thus, in this work, we aimed to systematically investigate the system where the S atom takes the place of the N atom. As shown in Fig. 1a, there are seven types of N atoms in the heptazine (C6N7) unit of melon-based CN, including 2-fold coordinated N (N1 to N6) and 3-fold coordinated N (N7). The optimized structures of the N1 site S-doped melon-based CN are shown in Fig. 1b and the optimized configurations of six other doping systems are displayed in Fig. S1 (ESI). Since the atomic radius of S is much larger than that of N, the S atom in the doped configuration is out-of-the-plane after heteroatom embedment. To determine the thermodynamic stability of doping systems substituting the S atom with the N atom, the formation energies (Eform)29,37 were calculated as follows:
Eform = E(S-melon) − E(melon) − μ(S) + μ(N)
where E(S-melon) and E(melon) are the total energies of the S-doped melon-based CN and the melon-based CN, respectively, and μ(S) and μ(N) are the chemical potentials of a single S atom and N atom, respectively. Here, μ(N) was set to be μ(N2)/2, as described in ref. 37.

image file: d0cp03533e-f1.tif
Fig. 1 (a) Top view of the possible doping sites (N1, N2, N3, N4, N5, N6, N7). (b) Top and side views of the geometric structure of the N1 site S-doped melon-based CN. The brown, gray, light pink and yellow balls represent C, N, H and S atoms, respectively.

The calculated formation energies are listed in Table 1. The results indicate that the substitution of the 3-fold coordinated N (N7) with S is the least thermodynamically favorable, which is in accordance with previous work covered by Liu et al.,35 while the thermodynamic stabilities of the other six S-doped systems were similar. As a result, the S-doped systems, except for the N7 doped site were chosen to systematically investigate the photocatalytic ORR in this work.

Table 1 The formation energies (Eform) of different N-site doping systems
Structure N1 N2 N3 N4 N5 N6 N7
E form (eV) −1.19 −1.40 −1.27 −1.07 −1.36 −1.19 −0.51

3.2. Electronic properties of S-doped melon-based CN

To gain insight into the dopant-induced changes in electronic properties, the band structure and PDOS were calculated. As displayed in Fig. 2a, melon-based CN is a semiconductor at the Γ-point, with a direct bandgap of 2.49 eV, which is slightly less than the experimental value (2.70 eV).18 However, it is well-established that the PBE functional generally underestimates the bandgap of semiconductors relative to the experimental value and it is challenging to calculate the bandgap of carbon nitride.52 When N is replaced by S, the bandgap is slightly reduced as compared to that of the melon-based CN (Fig. 2), which coincides with a recent report.38 It is worth noting that the band structures have different spin-up and spin-down states, suggesting that the samples were transformed into spin polarization systems after S doping due to the uneven valence electrons between the N and S atoms. Accordingly, a localized doping state appeared in the spin-up state between the conduction band minimum (CBM) and the valence band maximum (VBM), which indicated that the impurity level created by the S impurities in the PCN structure was observed experimentally.38 Note that S-doped melon-based CN are n-type semiconductors as the impurity band introduced by dopants causes the Fermi level to be near the bottom of the CB, and photoexcited electrons can easily transition from the VB to the doping state, and also from the doping state to the CB, which is rather favorable for O2 to capture electrons to get over kinetic obstacles in the ORR. The PDOS is displayed in Fig. 3; the VB edge is mainly composed of N atoms, and the CB edge consists of N and C atoms on the melon-based system. For S-doped melon-based CN, undoubtedly, a doping state was distinctly observed from the PDOS according to the band structure diagrams (Fig. 2). The doping state is mainly dominated by C, and N atoms, and a small amount of S atoms. Nevertheless, the composition of the CB and VB edges in the S-doped systems did not change significantly as compared to melon-based CN. The PDOS analysis is in accordance with the work derived by Chuaicham et al.38 The orbital diagrams of the VBM, CBM and the doping state of melon-based CN and S-doped systems are plotted in Fig. S2 (ESI). For melon-based CN, VBM is mainly clustered on N atoms, while CBM is distributed on N and C atoms, agreeing well with the PDOS analysis. The imino N (–NH) is neither a component of VBM nor a component of CBM, indicating that photoinitiated electrons are seriously confined to each heptazine unit, causing a high recombination rate of photogenerated carriers and poor photocatalytic efficiency. For S-doped systems, the orbital distributions of the CBM and VBM were notably converted due to the presence of the doping state. As displayed in Fig. S2 (ESI), the doping states of N3, N5, and N6 are mainly distributed on the C, N and S atoms on the heptazine unit around the S atom, while the doping states of N1, N2, and N4 are more delocalized on the heptazine unit chains. On the contrary, CBM is mainly composed of C, N atoms on the heptazine unit without S doping, unlike the pure melon-based CN. The VBM of N1, N2, N5, and N6 were mostly formed on the heptazine unit with S doping and the VBM of N3 and N4 were more spread out in the whole system. While the VBM, CBM, and doping state among S-doped systems are slightly different, the orbital distribution was altered observably by S atom embedment, which is more favorable for inhibiting the recombination of the photogenerated charge carriers and prefers the separation of photoexcited electron–hole pairs, improving the photocatalytic activity.
image file: d0cp03533e-f2.tif
Fig. 2 Band structure of (a) melon-based CN, (b) N1, (c) N2, (d) N3, (e) N4, (f) N5 and (g) the N6 site S-doped melon-based CN.

image file: d0cp03533e-f3.tif
Fig. 3 PDOS of (a) melon-based CN, (b) N1, (c) N2, (d) N3, (e) N4, (f) N5 and (g) the N6 site S-doped melon-based CN.

3.3. Optical properties of S-doped melon-based CN

Whether a photocatalyst can respond in the visible light range is of great importance. It is well known that the relatively large bandgap of melon-based CN restricts the adsorption of the broad solar illumination. Our previous work has shown that the performance of melon-based CN modified by S doping in the visible light region was effectively enhanced,39 and the fact is verified by the following calculated results. As plotted in Fig. 4, compared with melon-based CN, most of the optical adsorption edges of S-doped melon-based CN are red-shifted, which can be ascribed to the reduction of the bandgap, corresponding to the band structures (see Fig. 2). The red-shift in the absorption spectra to broaden the region of visible light in S-doped systems improve the ability to harvest the light with a longer wavelength, which results in the efficient generation of more electron–hole pairs and ultimately enhances the photoreactivity capacity. The Eform of N1 to N6 sites are similar, indicating that multiple doping sites can likely coexist. Hence the optical adsorption results indicate that the photocatalytic activities of S-doped melon-based CN should be greater than that of melon-based CN.
image file: d0cp03533e-f4.tif
Fig. 4 Optical absorption behaviors of melon-based CN and S-doped melon-based CN.

3.4. Catalytic performance for the selective ORR on S-doped melon-based CN

3.4.1. Activity origin of S-doped melon-based CN. The Bader charge of melon-based CN and S-doped melon-based CN were comparatively analyzed to uncover the origin of the distinct catalytic activity on doped samples, as displayed in Fig. S3 (ESI). Except for the striking influence on the charges of two C atoms adjacent to S, the charges of the atoms showed negligible differences, indicating that the excess electrons caused by the introduction of the S atom were localized and mainly dispersed on the two nearby C atoms. Note that one of the C atoms around the S atom accumulated the most electrons and the charge is indicated by the blue color in Fig. S3 (ESI). The corresponding C atoms for N1, N2, N3, N4, N5, N6 sites S-doped melon-based CN accumulated 0.72, 0.55, 0.56, 0.52, 0.47, and 0.79 e, respectively, predicting that the C atom with more electrons for the ORR would be the potential active site. To further confirm this point, we adsorbed O2 on all possible adsorption sites of catalyst surfaces to determine the most stable adsorption state, as well as calculated the electronic interactions between O2 and the catalyst surfaces under optimal adsorption configuration. On melon-based CN, O2 floats on the catalyst surface with a distance of approximately 3.00 Å (Fig. S4, ESI), leading to an ignorable electron transfer between O2 and the melon-based CN surface (Fig. 5). The adsorption energy was only −0.10 eV (Table 2), and it is a prerequisite for the ORR that O2 is adsorbed on the catalyst surface with chemisorption characteristics. Accordingly, the melon-based CN has the least attraction to O2, resulting in its lack of capacity to reduce O2. In stark contrast, on S-doped systems, one oxygen atom of O2 bonds with a C atom that is electron-rich, causing the reduction of the distances between O2 and the catalyst surfaces with the bond length of C–O ranging from 1.87 Å to 2.08 Å, and the O–O bond elongated to about 1.29 Å (vs. 1.23 Å on gaseous O2), as displayed in Fig. S4 (ESI). The charge transfer can be verified by the charge density difference of O2 adsorption on S-doped systems (Fig. 5); electron aggregation mainly occurs on the antibonding orbital of the adsorbed O2, while the depletion areas are around the C atom, as well as the O2 bonding orbital, which elongates the O–O bond. Correspondingly, the catalyst surface donates around 0.4 e to O2 in terms of Bader charge. Simultaneously, the electron transfer greatly enhances the O2 chemisorption on catalyst surfaces with the adsorption energies of −0.83 eV (N1), −0.55 eV (N2), −0.61 eV (N3), −0.71 eV (N4), −0.60 eV (N5), and −0.72 eV (N6), respectively (Table 2). These results indicate that the appropriate redistribution of charge density by doping with S heteroatoms can remarkably promote O2 adsorption.
image file: d0cp03533e-f5.tif
Fig. 5 Charge density difference profiles and Bader charges of O2 adsorption on (a) melon-based CN, (b) N1, (c) N2, (d) N3, (e) N4, (f) N5 and (g) N6 site S-doped melon-based CN. Charge density difference (Δρ) is defined as Δρ = ρ(O2–CN) − ρ(O2) − ρ(CN), where ρ(O2–CN), ρ(O2) and ρ(CN) are the charge densities of the O2-adsorbed system, isolated O2 and the S-doped melon-based CN system, respectively. Orange and cyan represent electron accumulation and depletion, respectively. The brown, gray, light pink, yellow and red balls represent C, N, H, S and O atoms, respectively.
Table 2 The adsorption energy (Eads) for adsorbates on melon-based CN and S-doped melon-based CN
Adsorbates E ads (eV)
Pure N1 N2 N3 N4 N5 N6
O2 −0.10 −0.83 −0.55 −0.61 −0.71 −0.60 −0.72
OOH −0.65 −2.41 −1.96 −1.86 −1.96 −1.81 −1.99
O −2.64 −4.59 −4.46 −4.99 −4.87 −4.61 −4.71
OH −0.63 −3.59 −3.32 −3.45 −3.42 −3.31 −3.48
H2O2 −0.51 −0.55 −0.48 −0.53 −0.54 −0.49 −0.52
H2O −0.36 −0.43 −0.38 −0.42 −0.31 −0.41 −0.33

3.4.2. The catalytic mechanism for the ORR. The ORR can occur via a 2e selective ORR procedure (eqn (1) and (2)) to form H2O2 or a 4e full ORR process (eqn (1) and (3)–(5)) to produce H2O. The elementary reaction steps are listed below:
* + O2 + H+ + e → *OOH(1)
*OOH + H+ + e → * + H2O2(2)
*OOH + H+ + e → *O + H2O(3)
*O + H+ + e → *OH(4)
*OH + H+ + e → * + H2O(5)
where * stands for an adsorption site on the catalysts, *OOH, *O, and *OH represent the oxygen-containing intermediates.

The 2e pathway leads to the direct synthesis of H2O2 by the two-step hydrogenation of the oxygen molecule, and only includes *OOH as the reaction intermediate, while the four-electron process undergoes a four-step reduction reaction to form H2O. According to the ORR mechanism, the adsorption energies of all relevant species on different catalyst surfaces during the reaction were calculated and are shown in Table 2. All possible adsorption sites for various species involved in the ORR process have been carefully considered, and the optimal adsorption configuration has been determined (Fig. S4, ESI). To gain insight into the dopant-induced reaction activity differences at the atomic level, we calculated ΔG for the ORR through the aforementioned mechanisms (Fig. 6).

image file: d0cp03533e-f6.tif
Fig. 6 The free energy diagrams of the ORR on (a) the melon-based CN, (b) N1, (c) N2, (d) N3, (e) N4, (f) N5 and (g) N6 site S-doped melon-based CN.

For melon-based CN, the adsorption characteristic of O2 is weak physisorption (−0.10 eV). Similarly, the adsorption energies of H2O2 and H2O were −0.51 and −0.36 eV, respectively, suggesting that the adsorption feature is between physisorption and weak chemisorption. The ΔG in the initial reduction step for ORR is downhill only by 0.05 eV as displayed in Fig. 6a, indicating that the step of protonation of O2 to form the *OOH intermediate is a challenging process due to the weak physisorption of O2 on the surface, as well as the slightly exothermic reaction to form *OOH.

The S-doped melon-based CN was successfully synthesized in the experiment.35–39,53–55 Interestingly, its application to the photocatalytic O2 reduction causing the formation of H2O2 was also experimentally determined.39 In this context, the O2 reduction process in the S-doped systems was discussed from atomic insight. The first step of the ORR is O2 adsorption on catalysts, and the adsorption feature of O2 is transformed into strong chemical adsorption on S-doped systems. In Section 3.4.1, we confirmed the fact that O2 was activated after S atom modification from several aspects, such as electron interactions, the distance between the oxygen molecule and the surface, as well as the oxygen molecule bond length, etc. Surprisingly, even if the polarity of melon-based CN was changed by S doping, there was little effect on the adsorption energy of H2O2 and H2O (Table 2). Moreover, the H2O2 and H2O surfaces maintained the reference structures. Hence, there is a high probability that H2O2 and H2O can desorb from the surface once they are produced on the catalyst surfaces. The values of ΔG for the 2e pathway are displayed in Table 3. Due to S doping, ΔG of the first protonation of O2 was remarkably reduced to −1.81 eV (N1), −1.34 eV (N2), −1.26 eV (N3), −1.36 eV (N4), −1.21 eV (N5) and −1.39 eV (N6), respectively, which can be attributed to the great activation of O2 by the decoration of S atom within the melon-based CN. In the generation of the H2O2 pathway, ΔG of *OOH protonation on the S-doped systems were downhill, except for the N1 site doping system, while the system is likely to be a mixed dopant in the experiment as stated above. Therefore, it can be appreciated that the 2e process of the second step of *OOH protonation to form H2O2 is an exothermic process. The overall exothermic process in the 2e pathway demonstrates that the synthesis of H2O2 on S-modified melon-based CN is thermodynamically feasible. Nevertheless, it is not neglected that the second protonation of the 4e mechanism is more dramatically downhill than the 2e pathway, indicating that it is more advantageous to form H2O than H2O2 from the viewpoint of thermodynamics. Judging from the U0 value (equilibrium potential for the reaction), the 4e process (1.23 VRHE) always emits more heat than the 2e pathway (0.70 VRHE).3,5 Consequently, this is an undoubted fact that the dissociation of *OOH generating H2O production is more favorable than H2O2 formation from a thermodynamics perspective. The calculated results indicate that S-doping would activate the O2 adsorption, while the dramatically enhanced activity of H2O2 production in PCN with controllable amounts of KCl and NaOH (sample named AKMT in ref. 39) might come from the combined effects of the doping with the KCl and NaOH during the thermal condensation, where further investigation is necessary. In all ORR catalysts, the selectivity of 2e must be affected by kinetic factors. Despite a great number of experiments dedicated to investigating the reaction kinetics of H2O2 selectivity, it is quite difficult to calculate precise energy barriers, leading to this controversy.4 In this work, the trend of the Gibbs free energy graphs are declining on the 2e mechanism as plotted in Fig. 7, suggesting that the generation of H2O2 catalyzed by S-doped systems is thermodynamically achievable as it is a spontaneous process, which is consistent with our previous work.39

Table 3 The values of ΔG for the 2e pathway on the melon-based CN and S-doped melon-based CN. ΔG1 and ΔG2 represent the free energy changes of the first step protonation and the second step protonation, respectively
Structure ΔG1 (eV) ΔG2 (eV)
Pure −0.05 −1.35
N1 −1.81 0.41
N2 −1.34 −0.06
N3 −1.26 −0.14
N4 −1.36 −0.04
N5 −1.21 −0.19
N6 −1.39 −0.01

image file: d0cp03533e-f7.tif
Fig. 7 The free energy diagrams of the 2e pathway on the melon-based CN and S-doped melon-based CN.

4. Conclusions

In this work, the photocatalytic (selective) ORR on S-doped melon-based CN has been systematically investigated with the DFT approach. The sulfur doping effects on the geometric configurations, electronic properties, optical properties, as well as the reaction performance of the ORR have been studied. Theoretical calculations demonstrate that the S atom embedment displayed evident and substantial effects over the corresponding photoreaction. The bandgap was slightly smaller as compared to the melon-based CN, resulting in the enhancement of visible light absorption. S-doped systems are n-type semiconductors owing to the introduction of the doping state after S-doping, leading to the Fermi level being near the CBM, and photoinduced electrons can readily jump from the VB to the doping state or from the doping state to the CB, which boosts the photoreduction ability. Consequently, O2 is well activated on the S-doped melon-based CN. The ΔG diagrams confirmed that the formation of H2O2 was thermodynamically achievable as the selective ORR process is exothermic on the S-doped melon-based CN, although we have to admit that the calculated model system is too idealized to mimic the experimental melon-based AKMT system in our previous work.39 Keeping this in mind, more realistic and complicated models will be built to further investigate the reaction. Furthermore, the trinary melon-based CN used in this work includes hydrogen bonds, which is structurally different from the mostly reported binary g-C3N4 (only contains C and N elements). However, most of the reported PCNs synthesized by the traditional thermally induced polymerization process are melon-based CNs instead of g-C3N4.56,57 Our work has revealed the potential nonmetal-doped impacts within the PCN in the photocatalytic ORR, providing favorable guidance for the design and development of PCN as well as other metal-free photocatalysts for the selective ORR.

Conflicts of interest

There are no conflicts to declare.


This work was financially supported by the National Natural Science Foundation of China (21973014, 21773030). W. L. thanks the fund for Minjiang Professorship of Fujian Province.


  1. R. Ciriminna, L. Albanese, F. Meneguzzo and M. Pagliaro, ChemSusChem, 2016, 9, 3374–3381 CrossRef CAS .
  2. J. M. Campos-Martin, G. Blanco-Brieva and J. L. G. Fierro, Angew. Chem., Int. Ed., 2006, 45, 6962–6984 CrossRef CAS .
  3. S. Yang, A. Verdaguer-Casadevall, L. Arnarson, L. Silvio, V. Colic, R. Frydendal, J. Rossmeisl, I. Chorkendorff and I. E. L. Stephens, ACS Catal., 2018, 8, 4064–4081 CrossRef CAS .
  4. V. J. Bukas, H. W. Kim, R. Sengpiel, K. Knudsen, J. Voss, B. D. McCloskey and A. C. Luntz, ACS Catal., 2018, 8, 11940–11951 CrossRef CAS .
  5. A. Kulkarni, S. Siahrostami, A. Patel and J. K. Nørskov, Chem. Rev., 2018, 118, 2302–2312 CrossRef CAS .
  6. Y. Shiraishi, S. Kanazawa, Y. Sugano, D. Tsukamoto, H. Sakamoto, S. Ichikawa and T. Hirai, ACS Catal., 2014, 4, 774–780 CrossRef CAS .
  7. Y. Shiraishi, S. Kanazawa, Y. Kofuji, H. Sakamoto, S. Ichikawa, S. Tanaka and T. Hirai, Angew. Chem., Int. Ed., 2014, 53, 13454–13459 CrossRef CAS .
  8. T. Hirai, ACS Catal., 2015, 5, 3058–3066 CrossRef .
  9. Y. Kofuji, Y. Isobe, Y. Shiraishi, H. Sakamoto, S. Tanaka, S. Ichikawa and T. Hirai, J. Am. Chem. Soc., 2016, 138, 10019–10025 CrossRef CAS .
  10. G.-h. Moon, M. Fujitsuka, S. Kim, T. Majima, X. Wang and W. Choi, ACS Catal., 2017, 7, 2886–2895 CrossRef CAS .
  11. Y. Zheng, Z. Yu, H. Ou, A. M. Asiri, Y. Chen and X. Wang, Adv. Funct. Mater., 2018, 28, 1705407 CrossRef .
  12. P. Zhang, D. Sun, A. Cho, S. Weon, S. Lee, J. Lee, J. W. Han, D. Kim and W. Choi, Nat. Commun., 2019, 10, 940 CrossRef .
  13. M. R. Hoffmann, S. T. Martin, W. Choi and D. W. Bahnemann, Chem. Rev., 1995, 95, 69–96 CrossRef CAS .
  14. E. Pelizzetti, Chem. Commun., 2005, 2627–2629 Search PubMed .
  15. M. N. Chong, B. Jin, C. W. K. Chow and C. Saint, Water Res., 2010, 44, 2997–3027 CrossRef CAS .
  16. D. Tsukamoto, A. Shiro, Y. Shiraishi, Y. Sugano, S. Ichikawa, S. Tanaka and T. Hirai, ACS Catal., 2012, 2, 599–603 CrossRef CAS .
  17. V. Diesen and M. Jonsson, J. Phys. Chem. C, 2014, 118, 10083–10087 CrossRef CAS .
  18. X. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M. Carlsson, K. Domen and M. Antonietti, Nat. Mater., 2009, 8, 76–80 CrossRef CAS .
  19. K. M. Parida, Inorg. Chem. Front., 2016, 3, 336–347 RSC .
  20. J. Wen, J. Xie, X. Chen and X. Li, Appl. Surf. Sci., 2017, 391, 72–123 CrossRef CAS .
  21. A. Nikokavoura and C. Trapalis, Appl. Surf. Sci., 2018, 430, 18–52 CrossRef CAS .
  22. S. Cao, J. Low, J. Yu and M. Jaroniec, Adv. Mater., 2015, 27, 2150–2176 CrossRef CAS .
  23. Y. J. Cui, Y. X. Wang, H. Wang and F. Y. Chen, Prog. Chem., 2016, 28, 428–437 CAS .
  24. T. Ye, D. P. Durkin, N. A. Banek, M. J. Wagner and D. Shuai, ACS Appl. Mater. Interfaces, 2017, 9, 27421–27426 CrossRef CAS .
  25. Z. Chen, J. Zhao, C. R. Cabrera and Z. Chen, Small Methods, 2019, 3, 1800368 CrossRef .
  26. Z. Guo, Y. Xie, J. Xiao, Z.-J. Zhao, Y. Wang, Z. Xu, Y. Zhang, L. Yin, H. Cao and J. Gong, J. Am. Chem. Soc., 2019, 141, 12005–12010 CrossRef CAS .
  27. H. Niu, X. Wang, C. Shao, Y. Liu, Z. Zhang and Y. Guo, J. Mater. Chem. A, 2020, 8, 6555–6563 RSC .
  28. S. Ye, R. Wang, M. Wu and Y. Yuan, Appl. Surf. Sci., 2015, 358, 15–27 CrossRef CAS .
  29. S. Lu, C. Li, H. H. Li, Y. F. Zhao, Y. Y. Gong, L. Y. Niu, X. J. Liu and T. Wang, Appl. Surf. Sci., 2017, 392, 966–974 CrossRef CAS .
  30. Z. Lin and X. Wang, Angew. Chem., Int. Ed., 2013, 52, 1735–1738 CrossRef CAS .
  31. Y. Li, S. Wang, W. Chang, L. Zhang, Z. Wu, R. Jin and Y. Xing, Appl. Catal., B, 2020, 274, 119116 CrossRef CAS .
  32. F. Wei, Y. Liu, H. Zhao, X. Ren, J. Liu, T. Hasan, L. Chen, Y. Li and B.-L. Su, Nanoscale, 2018, 10, 4515–4522 RSC .
  33. Y. Zhang, T. Mori, J. Ye and M. Antonietti, J. Am. Chem. Soc., 2010, 132, 6294–6295 CrossRef CAS .
  34. Y. Guan, J. Bai, S. Hu, P. Li, F. Wang, J. Ju and Y. Zhao, Diamond Relat. Mater., 2020, 105, 107799 CrossRef CAS .
  35. G. Liu, P. Niu, C. Sun, S. C. Smith, Z. Chen, G. Q. Lu and H. Cheng, J. Am. Chem. Soc., 2010, 132, 11642–11648 CrossRef CAS .
  36. K. Wang, Q. Li, B. Liu, B. Cheng, W. Ho and J. Yu, Appl. Catal., B, 2015, 176–177, 44–52 CAS .
  37. Y. Wang, Y. Tian, L. Yan and Z. Su, J. Phys. Chem. C, 2018, 122, 7712–7719 CrossRef CAS .
  38. C. Chuaicham, S. Karthikeyan, R. R. Pawar, Y. Xiong, I. Dabo, B. Ohtani, Y. Kim, J. T. Song, T. Ishihara and K. Sasaki, Chem. Commun., 2020, 56, 3793–3796 RSC .
  39. P. Zhang, Y. W. Tong, Y. Liu, J. J. M. Vequizo, H. W. Sun, C. Yang, A. Yamakata, F. T. Fan, W. Lin, X. C. Wang and W. Choi, Angew. Chem., Int. Ed., 2020, 59, 16209–16217 CAS .
  40. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186 CrossRef CAS .
  41. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775 CrossRef CAS .
  42. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS .
  43. K. Lee, E. Murray, L. Kong, B. I. Lundqvist and D. C. Langreth, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 081101 CrossRef .
  44. H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Condens. Matter Mater. Phys., 1976, 13, 5188–5192 CrossRef .
  45. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jonsson, J. Phys. Chem. B, 2004, 108, 17886–17892 CrossRef .
  46. J. Rossmeisl, A. Logadottir and J. K. Nørskov, Chem. Phys., 2005, 319, 178–184 CrossRef CAS .
  47. A. A. Peterson, F. Abildpedersen, F. Studt, J. Rossmeisl and J. K. Nørskov, Energy Environ. Sci., 2010, 3, 1311–1315 RSC .
  48. S. Zuluaga and S. Stolbov, J. Chem. Phys., 2011, 135, 134702 CrossRef .
  49. J. Rossmeisl, Z. W. Qu, H. Zhu, G. Kroes and J. K. Norskov, J. Electroanal. Chem., 2007, 607, 83–89 CrossRef CAS .
  50. X. Yu, X. Zhang, X. Tian, S. Wang and G. Feng, Appl. Surf. Sci., 2015, 324, 53–60 CrossRef CAS .
  51. M. D. Esrafili, F. Sharifi and L. Dinparast, J. Mol. Graphics Modell., 2017, 77, 143–152 CrossRef CAS .
  52. S. N. Steinmann, S. T. A. G. Melissen, T. Le Bahers and P. Sautet, J. Mater. Chem. A, 2017, 5, 5115–5122 RSC .
  53. Z. Pei, J. Gu, Y. Wang, Z. Tang, Z. Liu, Y. Huang, Y. Huang, J. Zhao, Z. Chen and C. Zhi, ACS Nano, 2017, 11, 6004–6014 CrossRef CAS .
  54. K. Chu, Q.-q. Li, Y.-p. Liu, J. Wang and Y.-h. Cheng, Appl. Catal., B, 2020, 267, 118693 CrossRef CAS .
  55. Y. Fang, X. Li, Y. Wang, C. Giordano and X. Wang, Appl. Catal., B, 2020, 268, 118398 CrossRef CAS .
  56. F. K. Kessler, Y. Zheng, D. Schwarz, C. Merschjann, W. Schnick, X. Wang and M. J. Bojdys, Nat. Rev. Mater., 2017, 2, 17030 CrossRef CAS .
  57. L. Lin, Z. Yu and X. Wang, Angew. Chem., Int. Ed., 2019, 58, 6164–6175 CrossRef CAS .


Electronic supplementary information (ESI) available. See DOI: 10.1039/d0cp03533e

This journal is © the Owner Societies 2020