Local charge polarization by introducing a cyanamide group and sulfur dopant with accelerated exciton dissociation and promotion of charge separation for improving CO₂ photoreduction performance

Abstract

Photocatalytic CO₂ reduction based on amorphous graphitic carbon nitride (GCN) has been severely restricted by the huge coulomb attraction between photoexcited electrons and holes (e⁻ and h⁺), as well as rapid recombination of photogenerated charge carriers. This exciton effect and sluggish charge separation dynamic originate from the symmetric electron cloud density. Herein, a two-step process (copolymerization and molten-salt treatment) with a heteroatom dopant and group introduction in crystalline GCN was developed to regulate charge distribution and construct local charge polarization (LCP). The as-prepared catalyst exhibited an improved CO₂ photoreduction performance with a CO product yield of 132.12 μmol g⁻¹ and CO selectivity of 95%. Experimental and theoretical calculations demonstrated that the cyanamide (–CN) group and a sulfur (S) dopant in heptazine conjugated ring led to the redistribution of photogenerated charge to form anisotropic charge distribution. These phenomena reduced the exciton binding energy and promoted exciton dissociation into free e⁻ and h⁺. Furthermore, a new lowest unoccupied molecular orbital (LUMO) was created that was lower than the LUMO state of unhybridized heptazine units, which accelerated charge carrier separation and extended the optical absorption range, finally improving the CO₂ photoreduction performance. This work presents an effective strategy to construct LCP in a catalyst via regulating electric and crystalline structures.

---

Yanrui Li

Yanrui Li is an Associate Professor in Xi'an University of Science and Technology. Dr Li received her doctorate from the University of Science and Technology of China, followed by a postdoctoral fellowship at Xi'an Jiaotong University. Her research focuses on exploring multifield coupling control over catalytic reactions, and developing energy coupling mechanisms for molecule transformations. This research is based on a high-performance photocatalyst for CO2 conversion into high value-added products and water-splitting into hydrogen. Her works have been published in Advanced Materials, Chemical Science, Advanced Functional Materials, Journal of Materials Chemistry A, and Small.

1. Introduction

Photocatalytic CO₂ reduction denotes conversion of a representative greenhouse gas into high value-added products, triggered by renewable solar-irradiation energy. This process offers a prominent solution for the corresponding environmental problems and addresses the energy crisis.^1–3 Graphitic carbon nitride (GCN) has excellent stability, low production cost and a suitable band structure for CO₂ reduction, and has been widely utilized for the photoreduction of CO₂. However, GCN (as an organic polymer) has been severely restricted by the huge coulomb attraction among photogenerated e⁻–h⁺ pairs, which originates from the symmetric electron cloud density. Meanwhile, the amorphous structure of GCN resulting from the low crystallinity brings about sluggish charge separation and transfer dynamics, which does not favour the performance of the photoreduction of CO₂.^4,5

Local charge polarization (LCP) engineering involves the redistribution of charge to form an inherent electric field. It has been effectively utilized to drive charge carrier separation and transport for inorganic semiconductors.^6–8 For organic polymers, the charge polarization effect not only accelerates the separation and migration of charge carriers but also acts as a strong driving force for exciton dissociation into free e⁻ and h⁺. This phenomenon should be taken into account for photocatalytic behavior based on organic polymers. For example, Li et al. varied ionic monomers to synthesize ionic covalent organic frameworks (iCOFs). They demonstrated that the ionic monomers had an orientated polarization in COF, which significantly endowed spontaneous exciton dissociation and further accelerated the separation and migration of free charge carriers.⁹ Zhu et al. utilized the polarization effects in organic solar cells to illustrate photoexcited charge generation mechanisms based on experimental and theoretical studies.¹⁰ The strong charge polarization effect enables organic solar cells to have a low exciton binding energy (E_b, −0.11 to 0.15 eV), favouring exciton dissociation into free e⁻ and h⁺. In general, the heptazine conjugated ring (the main component units for GCN) possesses a symmetric electron cloud density, which seriously confines e⁻–h⁺ pairs on the heptazine unit by attractive coulomb interactions.¹¹ Consequently, it is urgent to regulate the distribution of electrons on heptazine to create a local charge polarization environment. Such regulation would endow GCN a driving force for exciton dissociation into free photogenerated e⁻–h⁺ pairs and subsequently target charge carriers transfer to the active sites for the photoreduction of CO₂. For example, heteroatom dopants in heptazine conjugated rings could manipulate the electronic structure by the LCP to build an anisotropic charge separation, which favours a longer lifetime and trapped photogenerated electrons.¹² This LCP engineering exhibits outstanding capacity for selectively controlling exciton dissociation and charge migration for efficient photocatalytic production of H₂O₂. Furthermore, incorporating functional groups onto the heptazine conjugate framework, such as negatively charged COOH or OH groups,^7,13,14 has been shown to be an effective approach to enhance the π-system polarization of GCN. This strategy gives rise to functional groups with different polarity to heptazine conjugate rings, targets the redistribution of electron clouds, and then forms an inherent electric field. Consequently, breaking the symmetrical electron clouds of heptazine conjugate rings could be regarded as a promising strategy to construct LCP.

Apart from the electronic structure, the crystalline structure is also crucial for photocatalytic CO₂ reduction based on the impact of charge carrier migration. The amorphous structure of GCN (originally from the hydrogen bond-induced unordered arrangement of heptazine rings) leads to unidirectional migration and fast recombination of charge carriers.¹⁵ Therefore, improving the crystalline structure, constructing intraplanar and further interplanar ordered arrangement of heptazine conjugated rings, favours fast charge carrier transport and separation. The molten-salt method^16,17 provides an analogous liquid phase to endow the thermal polymerization of precursor molecules with adequate mass transport. This method has been regarded as an effective treatment to regulate the arrangement of heptazine units, as well as improve the crystallinity of GCN.^18,19

Based on the above reports on electric and crystal structure engineering for GCN, we developed a two-step process (copolymerization and molten-salt treatment) to introduce strong-polarity –CN groups and doped S heteroatoms in crystalline GCN. The as-prepared SCN-x exhibited enhanced performance for the photoreduction of CO₂, with CO product yield reaching up to 132.12 μmol g⁻¹ and CO selectivity of 95%. It was computationally demonstrated that the S dopants in heptazine rings and introduced –CN groups synergistically induced redistribution of the electron clouds of heptazine rings to form an anisotropic charge distribution, which could drive exciton dissociation and charge carrier transfer. Furthermore, dynamic characterizations confirmed that the recombination of photogenerated electron–hole pairs was effectively inhibited and the transport rate of charge carriers was dramatically strengthened thanks to heteroatom doping and the introduction of groups into crystalline GCN. Hence, LCP engineering by introducing groups and doping heteroatoms appears to be an effective strategy to restrain the exciton effect and strengthen charge separation dynamics for improving the performance of the photoreduction of CO₂ based on the GCN catalyst.

2. Experimental section

2.1 Synthesis of crystalline carbon nitride (CCN)

A mixture of as-prepared GCN powder and potassium thiocyanate (KSCN) in a mass ratio of 1 : 2 was placed in a tube furnace under an argon atmosphere. Then, this mixture was calcined under an argon atmosphere with the following heating program: (1) calcined at 140 °C for 8 h; (2) heated to 400 °C and held for 1 h; and (3) heated to 550 °C and held for 5 h. The heating rate remained at 5 °C min⁻¹ for all heating programs. After cooling down to room temperature, a light-yellow solid was obtained and washed several times with water to remove the residual salt.

2.2 Synthesis of sulfur-doped crystalline carbon nitride (SCN-x)

The pioneer powders containing sulfur dopant were first synthesized by calcining the mixture of melamine and thiourea at a mass ratio of 2 : 1, 1 : 1, 1 : 2 and 0 : 1 at 550 °C for 4 h with a heating rate of 5 °C min⁻¹ under an argon atmosphere (Scheme S1†). After cooling down to room temperature, the bulk was ground as the pioneer powders. Then, the pioneer powders were respectively mixed with KSCN at a mass ratio of 1 : 2 and heated using the same calcining procedure as that for the synthesis of CCN. According to different mass ratios of melamine and thiourea (2 : 1, 1 : 1, 1 : 2 and 0 : 1), the final products were named “SCN-2”, “SCN-1”, “SCN-0.5”, and “SCN” (SCN-x, x = 2, 1, 0.5 and 0; SCN-0 was abbreviated to “SCN”), respectively.

2.3 Photoreduction of CO₂

The performance of as-prepared catalysts for the photoreduction of CO₂ was evaluated in a gas–solid reactor with a 300 W xenon lamp (PLS-SXE300D; Perfect Light, China) with H₂O as a reducing agent. A quartz glass reactor (100 mL) with pre-filled 1 mL H₂O was equipped with a supported triangular scaffold to isolate the catalyst from H₂O. The catalyst (1 mg) was uniformly dispersed on the surface of the quartz sheet located on the supported triangular scaffold. After bubbling CO₂ gas for 30 min into the reactor, the gas–solid system was irradiated by the xenon lamp for 8 h. The photocatalytic reaction products were detected every 2 h by a gas chromatographs (GC 7820; GC, Huifen) equipped with a flame-ionization detector (FID) and thermal conductivity detector (TCD) via gas-tight syringes for intermittent sampling. The isotope-labelling products were detected by gas chromatography-mass spectrometry (GC-MS) using the 8860 system (Agilent Technologies). The apparent quantum yield (AQY) of CO evolution was evaluated under the same reaction conditions with different band-pass wavelength via the following equation:

3. Results and discussion

The crystalline structure of as-prepared catalysts was investigated using X-ray diffraction (XRD) and transmission electron microscopy (TEM). Two obvious diffraction peaks of GCN were located at 13.1° and 26.7° (Fig. S1†), which were indexed as (100) (d₁₀₀ = 6.75 Å) and (002) (d₀₀₂ = 3.23 Å) planes, which corresponded to in-plane periodicity and interlayer stacking of the aromatic conjugate ring,¹⁵ respectively. In comparison with GCN, the (002) peak of CCN displayed a significant shift from 26.7° to 27.5°, with the compressed interlayer distance changing from 0.323 nm to 0.321 nm (Table S1†), thereby favouring charge carrier transfer between interlayers. Additionally, the full width at maximum (FWHM) of the (002) plane declined from 1.42° to 0.820° (Table S1†), suggesting that molten-salt treatment favored enhanced crystallinity.²⁰ Based on the improved crystallization for CCN, the in-plane periodical units were extended with a shift in the in-plane periodicity (100) peak to 7.8° ((110), d₁₁₀ = 11.2 Å) and 9.8° ((020), d₀₂₀ = 8.9 Å).^21,22 Therefore, molten-salt treatment could effectively enhance the crystallinity with the compressed interlayer distance and extended in-plane periodical rings, which could be ascribed to the reassociation of the heptazine-based melon chains in the liquid environment supplied by the molten salt at a high temperature.²³ As illustrated in Fig. 1a, the XRD of SCN-x maintained the characteristic peaks of (020), (110) and (002) as CCN, demonstrating that the crystal structure and in-plane periodicity remained after doping S. Local magnification of the XRD patterns in Fig. 1 showed that, with an increase in the content of the S dopant, the characteristic peak of the (002) planes gradually shifted from 27.5° to 28.3°, along with the interlayer distance narrowing further from 0.321 to 0.315 nm (Table S1†), demonstrating strengthened interlayer interaction after S doping.²⁴ In addition, the FWHM of SCN-x increased compared with that of CCN (Table S1†), suggesting that S doping during self-polymerization affected the patterns and topology. However, the FWHM of SCN-x remained smaller than that of GCN, revealing that the improved crystallinity remained for SCN-x. The obvious lattice strips in Fig. 1b strongly confirmed that the long-range ordered crystalline structure was retained in SCN-x. In addition, the high-resolution TEM (HRTEM) image in the inset of Fig. 1b and integrated pixel intensities image along the (110) plane of SCN-0.5 (Fig. 1c) displayed a d-spacing of ∼11.5 Å, which approximated the interplanar spacing of the (110) plane calculated by XRD.

Fig. 1 (a) XRD patterns of CCN and SCN-x and local enlargement of the (002) peak (inset). (b) TEM image and HR-TEM image of SCN-0.5. (c) Integrated pixel intensities of SCN-0.5 along the (110) spacing direction. (d) FT-IR spectra of CCN and SCN-x. (e) C 1s and K 2p, (f) N 1s, and (g) S 2p XPS spectra of SCN-0.5. (h) Solid-state ¹³C NMR spectra of SCN-0.5. (i) Formation energies of SCN with the S atom replacing different triangular edge nitrogen (N1).

The chemical structure of CCN and GCN was revealed by Fourier transform infrared (FTIR) spectroscopy (Fig. S2†). Characteristic fingerprint bands at 1147–1642 cm⁻¹ and 810 cm⁻¹ for the vibration modes of conjugated heptazine units and flexural vibration of the heptazine ring could be discovered from CCN and GCN.^15,25 Significantly, two distinct changes in chemical bonds were discovered for CCN in comparison with GCN. A new band at 2180 cm⁻¹, ascribed to the telescopic vibration of –CN groups, could be attributed to the partial translation of amino groups into –CN groups by molten-salt treatment of KSCN. A new band at 993 cm⁻¹, indexed to the asymmetric vibration of C–N–C bonds in metal-NC2 groups, was ascribed to the introduced K⁺ to balance the charge of negative-charged –CN groups.^18,21 Additionally, the group replacement resulted in weakening of the band at 3140 cm⁻¹ indexed to the stretching vibration of amino groups, along with strengthening of the band at 3420 cm⁻¹ ascribed to hydrogen bonds,⁷ which could have because the negative-charged –CN groups preferred to form hydrogen bonds.²⁶ All the characteristic bands of CCN could also be observed from the FT-IR spectrum of SCN-x, suggesting that S doping had no effect on the skeleton structure of SCN-x. However, for SCN with only thiourea as a precursor, the band intensity of –CN increased compared with that of the other samples, which may have been due to the incomplete polymerization of the heptazine ring impacted by S doping.¹⁹ Hence, we can conjecture that the S atom may be present in the heptazine ring. Furthermore, the hydrogen bond band at 3420 cm⁻¹ in Fig. 1d gradually enhanced with an increase in content of the S dopant because the introduced S in heptazine rings with a strong polarity preferred to form hydrogen bonds with adjacent amino groups. To further verify the introduction of the S dopant, SEM and corresponding elemental mappings were performed (Fig. S3†). Significantly, the well-distributed S atoms integrated with the C, N and O atoms among SCN-0.5, which could be direct evidence for successful Si doping into SCN-x.

To accurately ascertain the atomical site of the S dopant in the heptazine conjugate ring and explore the corresponding polarization on the electron distribution of samples, X-ray photoelectron spectroscopy, density-functional theory (DFT) calculations and ¹³C solid-state nuclear magnetic resonance (NMR) spectroscopy were carried out. Except for two peaks centered at 284.8 eV and 288.0 eV ascribed to graphitic carbon and sp² hybridized carbon (N–C=N),²⁷ respectively, an additional peak located at about 286.6 eV, corresponding to –CN groups, was detected from CCN and SCN-x (Fig. 1e and S4a†). These data again confirmed the translation of partial –NH₂ groups into –CN groups during molten-salt treatment.¹⁸ Two K 2p peaks at 295.1 and 292.3 eV suggested the existence of K⁺ in SCN-x after molten-salt treatment, which was attributed to charge balance with the negative-charged –CN groups.²¹ Similar to GCN, four peaks centered at 398.1, 399.6, 400.7 and 404.1 eV were discovered in the N 1s spectra of CCN and SCN-x (Fig. 1f, S4b and c†), which were ascribed to the triangular edge nitrogen (N1), central tertiary nitrogen (N2), amino groups (N3), and nitrogen in the –CN groups (N4),^11,28–30 respectively. In addition, slight red-shifts for N1 were detected over SCN-x compared with GCN, which was likely due to reduced electron cloud density triggered by the electron-withdrawing –CN groups.¹⁸ This polarization process induced by –CN groups drove charge carrier redistribution and then electrons accumulated on the electron-withdrawing –CN groups, which could make the –CN groups preferential candidates for active sites for the photoreduction reaction. Furthermore, in the S 2p XPS spectra, the distinct peak centered at about 162.8 eV ascribed to S–C bonds could be fitted from SCN-x (Fig. 1g and S4d†), which confirmed that the S dopant had been introduced in the form of a S–C bond in heptazine conjugated rings.^12,31 With an increase in the content of thiourea, the C/S decreased, along with an increase in C/N for SCN-x (Table S2†), which strongly suggested introduction of the S dopant via replacing the N atoms in the heptazine ring. Therefore, we could confirm that S atoms, replacing the triangular edge nitrogen, formed the C–S–C bond in the heptazine conjugated ring. Additionally, the S dopant we introduced possessed a different electronegativity compared with that of the N and C atoms in the heptazine rings, which probably contributed to the charge carrier redistribution in the heptazine rings.³²

The –CN group at the terminus of the heptazine ring could induce the symmetric electron to be broken and redistributed in the heptazine rings. Hence, it was also possible for S dopant atoms with different electronegativity to that of C and N atoms to affect the electron distribution in heptazine rings. Therefore, it was necessary to accurately determine the replacement position of the S dopant in heptazine rings. Except for the two resonances at about 158 ppm (C1) and 166 ppm (C2), assigned to C atoms in the C–N₃ and CN₂–NH_x groups¹⁵ in both GCN, CCN and SCN-x, two new peaks at about 122 ppm (C3) and 171 ppm (C4) for CCN and SCN-x (Fig. 1h and S5†) were detected and ascribed to the C atoms in the –CN groups and C atoms adjacent to the –CN groups,²¹ respectively. Significantly, with an increase in S doping content, the resonance peak of C4 shifted from 172 ppm for CCN to 170 ppm for SCN-x. This was likely due to a strengthened shielding effect via increasing the density of the electron cloud by the adjacent S atom. This phenomenon confirmed that S dopant atoms replaced the N1 atoms adjacent to –CN groups. However, the ratios of C3 and C4 to C1 and C2 for SCN were critically different to those for SCN-0.5, SCN-1 and SCN-2, which was likely due to excess S dopant resulting in the disintegration of partial heptazine conjugated rings.^15,19 Furthermore, six models of N1 replaced by S dopant atoms were calculated and evaluated by the DFT calculation in Fig. S6† and 1i. Obviously, the formation energy of SCN-a displayed the most negative value compared with that of the other replacement models, which strongly demonstrated that S dopant atoms were likely to replace N1 adjacent –CN groups and K⁺.

To investigate the influence of LCP engineering by introducing the –CN group and S dopant on electron redistribution in heptazine rings and subsequent exciton dissociation, the DFT calculation and temperature-dependent PL spectroscopy were utilized. The electron density isosurface of CCN and SCN calculated by electron location function (ELF) revealed that the introduced –CN group and S dopant possessed a highly delocalized electron distribution (Fig. 2a–c) compared with that of GCN. Furthermore, calculation of the charge density difference confirmed electron enrichment on the –CN group (isosurface) and charge depletion on S dopant (isosurface) (Fig. S7†). For further exploration of the above phenomena, the band charge distributions were calculated and analyzed. The N atom of the –CN group enriched −0.48 electrons for CCN and −0.5 electrons for SCN, which made the –CN group a preferential candidate for the active sites for CO₂ conversion. In addition, in contrast with GCN, the electron density isosurface of SCN and CCN revealed an increase on N1 atoms adjacent to K⁺. This occurred because the coulomb force between positive K⁺ and N1 with a lone pair of electrons endowed more electron enrichment on N1 atoms adjacent to K⁺ over CCN and SCN, accelerating the polarization effect for heptazine units. Significantly, it was noticeable that more electrons enriched on N1 atoms adjacent to the K⁺ of SCN than that of CCN, along with a different electron distribution on the –CN group over SCN and CCN. This finding confirmed that the S dopant atoms acted as electronic donors for the heptazine rings increasing the electron density of the –CN group and N1 atom, further strengthening the polarization effect for heptazine units. Therefore, the S dopant along with introducing the –CN group synergistically regulated the charge redistribution and construction of asymmetrical electron enrichment and consumption regions in heptazine rings, leading to the generation of a local polarized electric field, which could provide a driving force for exciton dissociation. As shown in Fig. 2d, SCN-x possessed a negative electrostatic potential (ESP) on the –CN group and positive ESP on the S dopant with a bigger ESP difference than CCN and GCN, which favoured exciton dissociation during the photocatalytic process. This result could be further proved by the exciton binding energy (E_b) calculated from the temperature-dependent PL plots via the Arrhenius equation³³ (details in ESI†). As illustrated in Fig. 2e, upon introduction of the –CN group, the E_b reduced from 100.9 meV for GCN to 86.9 meV for CCN, demonstrating accelerated exciton dissociation driven by the LCP engineering of the –CN group. In addition, the S dopant could further reduce the E_b, attributed to an augmented charge difference as well as the corresponding ESP difference. However, upon increasing the S doping content, SCN-x exhibited a gradual increase in E_b and SCN-0.5 displayed the minimum value, suggesting that an S content of 0.5 favoured effective exciton dissociation.

Fig. 2 Electron density isosurface of (a) GCN, (b) CCN and (c) SCN. (d) Electrostatic potential (ESP) distribution of GCN, CCN, SCN and CO₂. (e) Temperature-dependent PL spectra of samples (λ_exciton = 370 nm). (f) Transient photocurrent action spectra and (g) photoluminescence spectra (PL) of samples GCN, CCN and SCN-x.

We recognized that LCP engineering by introducing a –CN group and S dopant could effectively strengthen exciton-dissociation behavior, meaning the generation of more free e⁻ and h⁺.³⁴ Hence, the corresponding charge separation and migration dynamics were investigated by photocurrent measurement and time-resolved photoluminescence (PL) emission spectroscopy. CCN exhibited a stronger photocurrent response than GCN, with the generation of more free e⁻ and h⁺ and effective separation due to the introduced –CN group and improved crystalline structure.³⁵ Additionally, with the S dopant and increasing the content of S, the photocurrent intensity of SCN-x continuously strengthened (Fig. 2f). This phenomenon was related to the S dopant further enhancing the LCP effect on exciton dissociation and endowing CCN with promoted charge separation efficiency. However, excess S dopant would result in charge recombination, along with weakened photocurrent intensity. The enhanced charge separation could be further confirmed by photoluminescence (PL), which originated from charge carrier recombination.¹¹ As shown in Fig. 2g, a distinct decrease was observed from GCN to CCN, representing the inhibited charge recombination by the improved crystalline structure and introduced –CN groups. Meanwhile, the PL intensity of SCN decreased further with S doping, accounted for by LCP inducing charge separation and then restraining charge recombination. Besides, the PL emission peak exhibited a slight red-shift from 460 nm for GCN to 410 nm for SCN, which corresponded to the narrowed bandgap. Similarly, with further increase in the content of doped S, the PL emission intensity increased. These photocurrent and PL analyses verified that use of SCN-0.5 should obtain optimization of the driving force for charge separation and migration, which was favourable for the photocatalytic performance for CO₂ conversion.

UV-visible diffuse reflectance spectroscopy (DRS) and DFT calculations were performed to further investigate the optical absorption ability, band structure, and electronic structure. In contrast to GCN, CCN exhibited a red-shift absorption edge from 461 nm for GCN to 490 nm for CCN (Fig. 3a), along with a narrower bandgap of 2.66 eV than GCN (2.82 eV), as calculated from the Kubelka–Munk function^36,37 (Fig. 3b). This narrowed bandgap for CCN could be ascribed to introducing –CN groups by replacing some amino groups, which was confirmed by DFT calculations. As illustrated in Fig. 3c and S8,† the highest occupied molecular orbital (HOMO) (N 2p orbitals) of CCN had the same energy as that of GCN, while the lowest unoccupied molecular orbital (LUMO) of CCN (hybridized by C 2p and N2p orbitals) was lower than that of GCN, which narrowed the bandgap of CCN. Light absorption intensity at 490–620 nm was distinctly enhanced, which was attributed to the translation of –CN groups with the n–π* electronic transition model.1,4,11 After introducing the S dopant in heptazine conjugate rings, acting as the main source of optical absorption units for a heptazine-based polymer,10,19 the optical absorption intensity of SCN-x gradually strengthened at 480–650 nm, as well as red-shift adsorption edges with narrowed bandgaps of 2.65, 2.64, 2.64 and 2.42 eV for SCN-0.5, SCN-1, SCN-2, and SCN, respectively. Additionally, except for bandgaps originating from a band-to-band transition, another bandgap at about 2.3 eV was detected for SCN-x (Fig. 3b), suggesting that a different charge transfer process occurred with introduction of the S dopant. DFT calculations demonstrated that a new LUMO state hybridized by the 2p states of C, N and S atoms in SCN (namely, hybridized-CNS unit) was generated that was lower than the LUMO of pristine heptazine units (unhybridized heptazine units). Comparison of the results of FTIR spectroscopy and XPS revealed that only some heptazine units were translated into hybridized-CNS units. The bandgap of the hybridized-CNS unit was lower than that of unhybridized heptazine units, which was contributed by the visible light absorption for SCN. Importantly, the new LUMO state could endow SCN-x a strong driving force to accelerate the separation of free charge carriers originating from the exciton dissociation. Hence, photoexcited electrons originating from the LUMO of unhybridized heptazine units would be trapped by the low-energy LUMO state of hybridized-CNS units. This DFT calculation was consistent with the charge carrier separation and migration dynamics results demonstrated by PL and photocurrent measurements. Combining the obtained conduction band (CB) minimum measured from Mott–Schottky (MS) plots (Fig. S9†), the band structures of as-prepared samples were schemed (Fig. 3e and S10†). The CB positions of all samples were sufficiently thermodynamically negative to drive the photocatalysis of CO₂ to CO.

Fig. 3 (a) UV-vis DRS spectra and (b) Tauc plots of GCN, CCN and SCN-x translated via the Kubelka–Munk function. DFT calculated band structure (left) and corresponding density of states (right) for (c) CCN and (d) SCN. (e) Band structure diagrams for an unhybridized heptazine unit and hybridized-CNS unit.

The photocatalytic performance for CO₂ conversion with H₂O vapor acting as the reducing agent was evaluated under the gas–solid model with the 300 W xenon lamp (λ > 420 nm) as the simulated solar light. As illustrated in Fig. 4a, CO and CH₄ were the main products for all as-prepared catalysts. CCN exhibited distinctly enhanced CO and CH₄ product activities in contrast with GCN, and the photoreduction CO₂ activity of SCN-x was obviously improved upon introduction of the S dopant. These results were consistent with the analysis of LCP engineering exciton dissociation and charge separation dynamics. In addition, excess doped S resulted in reduced activity due to critical charge recombination at the sites of excess doped S.³⁸ Hence, the stoichiometric S dopant for SCN-0.5 was suitable for the high-efficiency performance of CO₂ conversion, with the fastest CO product yield reaching up to 132.12 μmol g⁻¹ and CO selectivity of 95% (Fig. 4b). Compared with reported GCN-based photocatalysts without transition-metal or noble-metal modification, SCN-0.5 exhibited a superior CO₂-to-CO reaction rate and CO selectivity in the gas–solid reaction model without the need for a sacrificial agent. In addition, direct O₂ detection was hindered because the gas-tight syringes used for intermittent sampling had unavoidable air leakage, which resulted in artificially high O₂ background levels. Significantly, the photocatalytic activity of SCN-0.5 exhibited a slight reduction after 32 h of recycling (Fig. S11†), indicating high stability in the gas–solid reaction model. The AQY for SCN-0.5 under different types of monochromatic light irradiation was consistent with the optical absorption trend (Fig. 4c), indicating that the photocatalytic performance of CO₂ conversion was driven by light irradiation. Moreover, isotope labelling measurement with ¹³CO₂ gas illustrated ¹³CO (m/z = 29) to be the dominant reduction product detected by GC-MS (Fig. S12†). Hence, the carbon source for conversion of CO₂-to-CO originated from CO₂ rather than other carbon contaminates.

Fig. 4 (a) Time-dependent production for GCN, CCN and SCN-x under the gas–solid model. (b) Evolution rates of products for GCN, CCN and SCN-x, and the corresponding selectivity of CO. (c) Wavelength-dependent AQY and UV-visible DRS spectrum over SCN-0.5.

Bearing the above discussion in mind, LCP engineering by introducing a –CN group and S dopant induced an asymmetric charge distribution, which could effectively strengthen the exciton dissociation into free charge carriers. Meanwhile, the new LUMO state of hybridized-CNS units could effectively trap the free electrons from the HOMO of unhybridized heptazine units, helping to accelerate charge carrier separation and migration. Generally, the strengthened exciton dissociation and promoted charge separation dynamics could synergistically enhance the photocatalytic CO₂ reduction performance. Therefore, the correlative reaction process and mechanism were further investigated by in situ attenuated total reflectance (ATR)-FTIR spectroscopy and DFT calculations. As shown in Fig. 5a, with prolonged irradiation from 0 to 60 min, three new bands at 1593, 1539 and 1509 cm⁻¹, indexed to COOH* intermediates,^39–42 were generated and the intensity gradually strengthened, which suggested that the reaction process of CO₂-to-CO conversion involved a hydrogenation and deoxygenation reaction (CO₂(g) + e⁻ + H⁺ → COOH*, COOH* + e⁻ + H⁺ → H₂O + CO*).⁴³ As discussed previously, photoexcited electrons can accumulate on –CN groups induced by LCP engineering, making the –CN group a preferential active site for the photoreduction of CO₂. The charge-density difference of CO₂ adsorption on N1 for GCN and –CN groups for CCN and SCN was calculated and illustrated in Fig. 5b. The calculation illustrated an order of interaction strength between the catalyst and CO₂ molecule: SCN > CCN > GCN. Hence, electron-accumulated –CN groups were favorable for CO₂ adsorption and, following S dopant-induced charge distribution, could further enhance the electron density of the –CN group. As illustrated in Fig. S13,† the CO₂ adsorption curve for GCN exhibited a slight rise compared with that for CCN. Additionally, the curves for SCN-x displayed significant increase compared with GCN and CCN after introducing the S dopant. These data were consistent with the calculation results and confirmed a stronger ability of CO₂ adsorption for SCN-x than GCN and CCN. It is recognized that CO₂ adsorption, acting as the first step for the photoreduction of CO₂, initiates the subsequent multielectron conversion reaction.⁴⁴ To investigate the reaction mechanism for CO₂-to-CO conversion, the Gibbs free energy over GCN, CCN and SCN with each hydrogenation and deoxygenation step were simulated by DFT calculations. As shown in Fig. 5c, SCN possessed the lowest free energy difference of adsorbing compared with CCN and GCN. Meanwhile, generating COOH* displayed the maximum free energy difference (i.e., the rate-limiting step for CO₂-to-CO conversion).⁴⁵ A free energy difference of 1.42 eV required for generating COOH* over SCN was significantly reduced compared with that over GCN (1.82) eV and CCN (1.50 eV), which confirmed that LCP and S dopant engineering electron distribution were favourable for the hydrogenation of into the COOH* intermediate. Furthermore, a lowest free energy difference for the CO* formation required by SCN demonstrated that the CO* intermediate preferentially desorbed and generated CO gas, which explained the enhanced CO₂ photoreduction performance.

Fig. 5 In situ (ATR)-FTIR spectra for SCN-0.5. (b) Charge density difference of surface CO₂-adsorption on GCN, CCN and SCN. (c) DFT calculation of Gibbs free energy of the CO₂ reaction on GCN, CCN and SCN.

4. Conclusions

We introduced the –CN group and S dopant in heptazine rings via copolymerization and molten-salt treatment to regulate the electric and crystalline structure. This strategy accelerated exciton dissociation and charge carrier separation, and improved the CO₂ photoreduction performance. The obtained SCN-0.5 exhibited enhanced CO₂ photoreduction activity with a CO yield reaching up to 132.12 μmol g⁻¹, which was 7.4-times that using CCN. XPS, NMR spectroscopy and DFT calculations revealed the sites of the S dopant and –CN groups: replacement of sp² hybridized nitride in the heptazine ring and replacement of the terminal –NH₂ of the adjacent heptazine ring. Furthermore, theoretical calculations and temperature-dependent PL demonstrated that the S dopant and –CN groups in the heptazine ring regulated charge distribution and enhanced π-system polarization, reducing the E_b and accelerating exciton dissociation into free charge carriers. Additionally, experiments and calculations revealed that the photogenerated charge carriers targeted separation and migration to a new LUMO state comprised by a hybridized-CNS unit, which had a lower LUMO state of unhybridized heptazine units, which extended the optical absorption range. This work provides a reference for synergistic regulation of the electric and crystalline structures of catalysts to construct LCP for outstanding photocatalytic CO₂ reduction performance.

Data availability

The data supporting the conclusions reached in our study are available upon reasonable request from the corresponding author.

Author contributions

Yanrui Li: conceptualization, supervision, project administration, funding acquisition, and writing (review and editing). Bozhan Li: investigation, data curation, data analysis, resources, and investigation. Xiang Gao: supervision, funding acquisition, and visualization. Linda Wang: data analysis and investigation. Xuehao Li: data analysis and methodology. Ruyu Guo: data analysis and validation.

Conflicts of interest

There are no conflicts of interest to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (22303064), Outstanding Youth for Xi'an University of Science and Technology (2023YQ3-07), and Xi'an Association for Science and Technology Young Talent Support Program (959202413073).

References

1. H. Zhang, A. Yohannes, H. Zhao, Z. Li, Y. Xiao, X. Cheng, H. Wang, Z. Li, S. Siahrostami, M. G. Kibria and J. Hu, Nat. Commun., 2025, 16, 534.
2. J. Ding, P. Du, P. Li, W. Liu, J. Xu, W. Yan, Y. Pan, J. Hu, J. Zhu, Q. Chen, X. Jiao and Y. Xie, Angew. Chem., Int. Ed., 2025, 64, e202414453.
3. C. Chen, C. Ye, X. Zhao, Y. Zhang, R. Li, Q. Zhang, H. Zhang and Y. Wu, Nat. Commun., 2024, 15, 7825.
4. R. Nakada, C. Zhang, J. Onodera, T. Tanaka, M. Okazaki, G. Zhang and K. Maeda, Sustainable Energy & Fuels, 2025, 9, 947–952.
5. T. Tanaka, C. Zhang, R. Nakada, J. Onodera, F. Ishiwari, M. Okazaki, A. Saeki, O. Ishitani, G. Zhang and K. Maeda, CCS Chem., 2024, 6, 2940–2949.
6. W. J. Zhong, M. Y. Hung, Y. T. Kuo, H. K. Tian, C. N. Tsai, C. J. Wu, Y. D. Lin, H. C. Yu, Y. G. Lin and J. J. Wu, Adv. Mater., 2024, 36, 2403228.
7. M. Song, H. Song, P. Chen, D.-H. Lan, F. Liu, C.-T. Au and S.-F. Yin, ACS Sustainable Chem. Eng., 2023, 11, 8096–8105.
8. J. Li, H. Jin, T. Qin, F. Liu, S. Wu and L. Feng, ACS Nano, 2024, 18, 4539–4550.
9. G. Li, P. Fu, Q. Yue, F. Ma, X. Zhao, S. Dong, X. Han, Y. Zhou and J. Wang, Chem. Catal., 2022, 2, 1734–1747.
10. L. Zhu, J. Zhang, Y. Guo, C. Yang, Y. Yi and Z. Wei, Angew. Chem., Int. Ed., 2021, 60, 15348–15353.
11. Y. Li, Y. Wang, C.-L. Dong, Y.-C. Huang, J. Chen, Z. Zhang, F. Meng, Q. Zhang, Y. Huangfu, D. Zhao, L. Gu and S. Shen, Chem. Sci., 2021, 12, 3633–3643.
12. P. Zhang, Y. Tong, Y. Liu, J. J. M. Vequizo, H. Sun, C. Yang, A. Yamakata, F. Fan, W. Lin, X. Wang and W. Choi, Angew. Chem., Int. Ed., 2020, 59, 16209–16217.
13. G. Hu, H. Song, J. Cao, Y. Liu, C. Fu and P. Chen, J. Environ. Chem. Eng., 2024, 12, 114095.
14. J. Wan, Y. Wang, J. Liu, R. Song, L. Liu, Y. Li, J. Li, J. Low, F. Fu and Y. Xiong, Adv. Mater., 2024, 36, 2405060.
15. V. W.-h. Lau, I. Moudrakovski, T. Botari, S. Weinberger, M. B. Mesch, V. Duppel, J. Senker, V. Blum and B. V. Lotsch, Nat. Commun., 2016, 7, 12165.
16. G. Zhang, Y. Xu, P. Zhang, C. He and H. Mi, Adv. Funct. Mater., 2024, 34, 2315116.
17. G. Zhang, W. Huang, Y. Xu, Y. Li, C. He, X. Ren, P. Zhang and H. Mi, Adv. Funct. Mater., 2023, 33, 2305935.
18. Y. Li, Y. Xue, X. Gao, L. Wang, X. Liu, Z. Wang and S. Shen, Adv. Funct. Mater., 2023, 34, 2312634.
19. Z. Yu, X. Yue, J. Fan and Q. Xiang, ACS Catal., 2022, 12, 6345–6358.
20. L. Lin, H. Ou, Y. Zhang and X. Wang, ACS Catal., 2016, 6, 3921–3931.
21. G. Zhang, Y. Xu, D. Yan, C. He, Y. Li, X. Ren, P. Zhang and H. Mi, ACS Catal., 2021, 11, 6995–7005.
22. G. Zhang, Y. Xu, C. He, P. Zhang and H. Mi, Appl. Catal., B, 2021, 283, 119636.
23. I. G.-R. D. Portehault, M. Baron, V. Musumeci, C. Aymonier, V. Rouchon and Y. L. Godec, Chem. Soc. Rev., 2022, 51, 4828–4866.
24. P. N. G. Liu, C. Sun, S. C. Smith, Z. Chen, G. Q. Lu and H. Cheng, J. Am. Chem. Soc., 2010, 132, 11642–11648.
25. J. Liang, W. Zhang, Z. Liu, Q. Song, Z. Zhu, Z. Guan, H. Wang, P. Zhang, J. Li, M. Zhou, C. Cao, H. Xu, Y. Lu, X. Meng, L. Song, P. K. Wong, Z. Jiang and C.-S. Lee, ACS Catal., 2022, 12, 12217–12226.
26. M. Zhou, H. Wang, R. Liu, Z. Liu, X. Xiao, W. Li, C. Gao, Z. Lu, Z. Jiang, W. Shi and Y. Xiong, Angew. Chem., Int. Ed., 2024, 63, 202407468.
27. Y. Xu, M. Fan, W. Yang, Y. Xiao, L. Zeng, X. Wu, Q. Xu, C. Su and Q. He, Adv. Mater., 2021, 33, 2101455.
28. G. Zou, Q. Wang, G. Ye, Z. Pan, S. Wang, M. Anpo and G. Zhang, Adv. Funct. Mater., 2024, 2420899.
29. Q. Wang, D. Zheng, Z. Pan, W. Xing, S. Wang, Y. Hou, M. Anpo and G. Zhang, Adv. Funct. Mater., 2025, 2501889.
30. Y. Hu, Z. Yang, D. Zheng, W. Xing and G. Zhang, Nanoscale, 2025, 17, 7856–7864.
31. J. M. Z. Liu, M. Hong and R. Sun, ACS Catal., 2023, 13, 2106–2117.
32. J. Tang, X. Li, Y. Ma, K. Wang, Z. Liu and Q. Zhang, Appl. Catal., B, 2023, 327, 122417.
33. G. Chen, Z. D. Zhang, Y. X. Liao, Z. Zhang and Y. Z. You, Small, 2021, 17, 2100698.
34. H. Tan, W. Si, W. Peng, X. Chen, X. Liu, Y. You, L. Wang, F. Hou and J. Liang, Nano Lett., 2023, 23, 10571–10578.
35. S. Zheng, P. Lu, A. M. Idris, B. Chen, X. Jiang, G. Jiang, J. Wang, S. Li and Z. Li, J. Mater. Chem. A, 2025, 13, 5346–5356.
36. Y. Ren, Y. Fu, N. Li, C. You, J. Huang, K. Huang, Z. Sun, J. Zhou, Y. Si, Y. Zhu, W. Chen, L. Duan and M. Liu, Nat. Commun., 2024, 15, 4675.
37. F. Xiong, H. Hu, X. Xue, M. Wu, J. Zhou, W. Zhang and R. Li, Water Res., 2024, 259, 121888.
38. X. Xu, L. Meng, J. Zhang, S. Yang, C. Sun, H. Li, J. Li and Y. Zhu, Angew. Chem., Int. Ed., 2024, 63, e202308597.
39. J. Wang, Y.-C. Huang, Y. Wang, H. Deng, Y. Shi, D. Wei, M. Li, C.-L. Dong, H. Jin, S. S. Mao and S. Shen, ACS Catal., 2023, 13, 2374–2385.
40. L. Shen, Z. Xie, L. Hou, J. Yang and Q. Li, Energy Fuels, 2022, 36, 11515–11523.
41. Z. Wang, J. Zhu, X. Zu, Y. Wu, S. Shang, P. Ling, P. Qiao, C. Liu, J. Hu, Y. Pan, J. Zhu, Y. Sun and Y. Xie, Angew. Chem., Int. Ed., 2022, 61, e202203249.
42. T. Wang, L. Chen, C. Chen, M. Huang, Y. Huang, S. Liu and B. Li, ACS Nano, 2022, 16, 2306–2318.
43. K. Yan, D. Wu, T. Wang, C. Chen, S. Liu, Y. Hu, C. Gao, H. Chen and B. Li, ACS Catal., 2023, 13, 2302–2312.
44. N. Li, J. Wang, G. Zhao, J. Du, Y. Li, Y. Bai, Z. Li and Y. Xiong, ACS Mater. Lett., 2024, 6, 999–1006.
45. Y. Chai, Y. Kong, M. Lin, W. Lin, J. Shen, J. Long, R. Yuan, W. Dai, X. Wang and Z. Zhang, Nat. Commun., 2023, 14, 6168.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta02281a

---