Activating Ionic Carbon Nitrides via Epitaxial Dyadic Coupling for Efficient Solar‐Driven H₂O₂ Generation

ABSTRACT

Solar hydrogen peroxide synthesis is a sustainable alternative to the anthraquinone process, yet its efficiency is limited by charge separation and transport in polymeric photocatalysts. Ionic carbon nitrides like poly(heptazine imide) (PHI) offer tunable structures and electronic properties. Herein, we report an epitaxial strategy to construct a different π‐conjugated CN species (C₁N₁) on potassium poly(heptazine imide) (KPHI), forming a dyadic charge‐transfer nanostructure with strong interfacial coupling. Structural and spectroscopic analyses reveal that epitaxial registry at the interface induces orbital hybridization and a built‐in electric dipole, driving directional electron transfer from KPHI to C₁N₁ while retaining holes within the PHI. The conduction band of the composite shifts upward, providing a thermodynamic benefit for the oxygen reduction reaction (ORR). Theoretical calculations confirm the thermodynamic driving force for electron transfer from KPHI to C₁N₁ and strong interfacial interactions. The optimized composite achieves an AQY of 63.9% at 410 nm without cocatalysts, approaching the optimum for stirred powder reactors. In‐situ FTIR detects ∙O₂ ⁻ and ∙OOH intermediates and further reveals interfacial water reconstruction with increased free water molecules that facilitate proton supply. Mechanistic studies confirm a two‐electron ORR pathway mediated by superoxide intermediates, promoted by the programmed charge‐transfer landscape.

Epitaxial integration of a C₁N₁ phase onto KPHI creates a dyadic charge‐transfer nanostructure with strong interfacial electronic coupling. This architecture drives directional electron transfer from KPHI to C₁N₁, enabling efficient cocatalyst‐free solar H₂O₂ production.

1. Introduction

Artificial photosynthesis is considered as a sustainable approach to harness solar energy for the sustainable production of value‐added, energy‐rich molecules [1]. Importantly, the ability to control charge carrier dynamics in semiconductors lies at the heart of photocatalysis and energy conversion [2]. While significant progress has been made in designing materials with high chemical stability, suitable band structures and light absorption properties, it is still important to improve how photogenerated electrons and holes are spatially separated, transported, and ultimately utilized [3, 4, 5].

Ionic carbon nitrides, such as poly(heptazine imide) (PHI), represent a unique class of polymeric semiconductors that bridge molecular and solid‐state systems [6, 7]. Their extended π‐conjugated frameworks with embedded ionic environments enable not only efficient light absorption but also the accumulation of photogenerated electrons, thereby facilitating multi‐electron redox chemistry [8]. However, the electronic structure of pristine PHI remains largely homogeneous, and introducing the ability to direct charge flow is therefore highly desirable [9]. The introduction of electronically coupled secondary phases that can act as charge‐transfer partners is a typical approach to realize better charge separation and direction [10, 11, 12, 13, 14, 15]. Covalent materials with extended π‐conjugation are particularly attractive in this regard, especially when stacked as 2D systems onto the PHI. Sensitization of semiconductors with carbon dots can be considered as a first step in this direction, but does not provide optimized electronic coupling [16, 17, 18].

Here, we present that epitaxial integration of a different carbon nitride phase onto ionic carbon nitrides can create a dyadic electronic architecture, in which interfacial orbital hybridization gives rise to new charge‐transfer (CT) states. This special dyadic configuration enables strong electronic coupling and spatially resolved charge separation, allowing electrons and holes to be selectively localized in distinct domains [19]. We realize this concept through the synthesis of a C₁N₁‐KPHI hybrid via condensation in salt melts, where guanine‐derived C₁N₁ domains grow epitaxially on KPHI. Structural and spectroscopic analyses reveal the formation of a coherent interface that promotes charge‐transfer‐type orbital hybridization. Band structure analysis shows that the conduction band of the composite shifts upward to a more negative position, which is thermodynamically favorable for the ORR process. DFT calculations confirm that the Fermi level of KPHI (−4.51 eV) is higher than that of C₁N₁ (−5.41 eV), providing the thermodynamic driving force for electron transfer from KPHI to C₁N₁, while density of states (DOS) analysis further confirms strong interfacial interactions. Photoexcited electrons are preferentially transferred from KPHI to C₁N₁, where they are stabilized and accumulated, while holes remain confined within the KPHI framework. This dyadic architecture suppresses recombination and enables directional charge transport across the interface. As a model reaction, photocatalytic H₂O₂ generation via oxygen reduction was analyzed, and high efficiencies exceeding 60% apparent quantum yield could be realized. The in‐situ FTIR spectroscopy not only detects key reaction intermediates (∙O₂ ⁻ and ∙OOH), confirming the superoxide‐mediated two‐electron ORR pathway, but also reveals interfacial water reconstruction with an increase in free water molecules, which enhances proton transfer kinetics. These findings offer direct evidence for the reaction mechanism and highlight the crucial role of interfacial water dynamics in promoting H₂O₂ generation. More broadly, this work establishes epitaxial dyadic interface engineering as a strategy to program the electronic structure of polymeric semiconductors, moving beyond conventional heterojunction design toward adaptive and functionally integrated materials.

2. Results and Discussion

2.1. Materials Characterization

The catalysts were synthesized via a straightforward one‐step thermal condensation approach, as illustrated in Figure 1a. Typically, 2.5 g of 5‐aminotetrazole was mixed with 12.5 g of a KCl/LiCl eutectic salt mixture (mass ratio 0.55:0.45), followed by the addition of a specific amount of guanine, which served as a C₁N₁ precursor. Interestingly, the two monomers do not react with each other but instead condense into separate phases. The mixture was then ball‐milled at 25 Hz for 5 min, transferred to a crucible, and thermally annealed at 600 °C for 4 h under a nitrogen atmosphere. The obtained yellowish‐brown powders were denoted as x%C₁N₁‐KPHI, where x indicates the weight percentage of guanine relative to 5‐aminotetrazole (5‐AT). For comparison, pristine KPHI and C₁N₁ were also synthesized under identical conditions in the absence of guanine and 5‐AT, respectively. The structural and chemical properties of the synthesized photocatalysts were characterized by x‐ray diffraction (XRD) and Fourier transform infrared (FTIR) spectroscopy. In the XRD analysis, pure KPHI exhibits two characteristic peaks at 8.1° and 28.1°, which correspond to the (110) and (002) crystal planes, respectively (Figure S1a) [20]. In comparison, C₁N₁ displays a single broad peak centered around 27.1°, assigned to the (002) crystal plane [21]. These distinct diffraction patterns confirm the successful synthesis of both KPHI and C₁N₁ structures. FTIR spectroscopy provides further insight into the structure of C₁N₁ (Figure S1b). The disappearance of characteristic vibration peaks from the guanine precursor following thermal treatment confirms its transformation into C₁N₁. A broad absorption band in the 1200–1600 cm⁻¹ range is attributed to stretching vibrations of C‐N heterocycles, indicative of aromatic nitrogen‐rich structures. Another broad band between 2800 and 3700 cm⁻¹ corresponds to N─H stretching and bending modes and/or physically adsorbed water, suggesting that condensation of guanine‐derived intermediates may be incomplete. Additionally, peaks at 1250 and 1582 cm⁻¹ signify the presence of C═C and C═N functional groups, consistent with partial graphitization and the formation of heteroaromatic frameworks [22].

FIGURE 1.

(a) Illustration for the synthesis process of x%C₁N₁‐KPHI with proposed structures. (b) XRD patterns and (c) FTIR spectra of KPHI and all x%C₁N₁‐KPHI. HR‐TEM images of (d,e) C₁N₁, with (f) enlarged area pattern, (g,h) KPHI, (i,j) 2%C₁N₁‐KPHI and (k,l) 5%C₁N₁‐KPHI.

The Raman spectrum of C₁N₁ shows a peak centered at 1001 cm⁻¹, which is assigned to the vibrational mode of the triazine ring (Figure S1c). Two main peaks at approximately 1354 and 1502 cm⁻¹ correspond to overlapping C─N and C─C bond vibrations within the triazine framework. Additionally, the peak at 2798 cm⁻¹ indicates higher‐order vibrations typical of stacked carbonaceous layers [23]. Thermogravimetric analysis (TGA) assesses the thermal stability of the starting materials. As shown in Figure S1d, 5‐AT, used for KPHI synthesis, decomposes around 200°C, whereas guanine decomposes at a higher temperature (∼ 450°C). This disparity suggests that the poly(heptazine) network likely occurs before C₁N₁ formation when both precursors are combined, which is advantageous as it minimizes the likelihood of C₁N₁ interfering with KPHI assembly. This proposed sequential formation is further supported by XRD data (Figure 1b), where all x%C₁N₁‐KPHI hybrid samples exhibit the characteristic diffraction peaks of KPHI, indicating that its core structure remains largely intact after incorporating C₁N₁. A minor shift of these peaks toward lower angles hints at possible interactions or integration between KPHI and C₁N₁. In Figure 1c, KPHI and its hybrid composites exhibit nearly identical vibrational features, reinforcing the structural integrity of KPHI upon C₁N₁ incorporation. A distinct peak at 801 cm⁻¹ corresponds to the out‐of‐plane breathing vibration of the heptazine ring, while vibration at 910 and 987 cm⁻¹ is linked to C─N bonds associated with K⁺ coordination to ─NC₂ units. The range between 1200 and 1700 cm⁻¹ contains multiple bands attributed to stretching and bending vibrations of heptazine derivatives. Broad absorption in the 3000–3600 cm⁻¹ range is attributed to terminal amino (─NH₂) groups and adsorbed water [24, 25]. Notably, the intensity of this broad band diminishes progressively with increasing C₁N₁ content, implying that C₁N₁ may react or interact with the terminal amino groups during hybrid formation.

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) analyses revealed the morphological evolution of the photocatalysts. As illustrated in Figure S2, pure C₁N₁ appeared as small fragments or oval‐shaped particles, while pure KPHI displayed a distinct rod‐like structure. Upon the formation of the hybrid C₁N₁‐KPHI materials, the nanorod architecture of KPHI was generally preserved. However, increasing amounts of fine particles were observed decorating the nanorods as the C₁N₁ content increased. Especially, the 10% and 20%C₁N₁‐KPHI catalysts showed extensive surface coverage by C₁N₁ particles. This increased surface coverage is likely responsible for the reduced crystallinity observed, in agreement with the XRD analysis. High‐resolution TEM imaging reveals that C₁N₁ possesses a fragmented or pebble‐like morphology, with magnified selected‐area images displaying a clear lattice spacing of 0.33 nm, corresponding to the (002) crystallographic planes (Figure 1d–f). Elemental mapping images confirm the uniform distribution of C, N, and K within the C₁N₁ structure, verifying successful synthesis (Figure S3). For the heterojunction systems, TEM images of pristine KPHI and the C₁N₁‐KPHI composites consistently display a rod‐like morphology. Notably, smaller fragments are observed surrounding the nanorods, and their presence becomes more pronounced with increasing C₁N₁ content, as illustrated in Figure S4. Further high‐resolution TEM analysis provides compelling evidence for the formation of a heterojunction between C₁N₁ and KPHI (Figure 1g–l). In comparison to pure KPHI, both 2%C₁N₁‐KPHI and 5%C₁N₁‐KPHI catalysts exhibit distinct lattice spacings of 1.06 and 0.31 nm, which correspond to the (110) and (002) planes, respectively. The regions marked with red dots in Figure 1j,l correspond to C₁N₁‐enriched domains. Notably, C₁N₁ appears to grow epitaxially along KPHI surfaces, exhibiting a lattice spacing of 0.33 nm that is consistent with the (002) planes of C₁N₁. The epitaxial growth is presumably driven by lattice matching along the stacking direction together with interfacial electron transfer. As the C₁N₁ content increases, more C₁N₁ domains are observed, confirming enhanced surface coverage in higher‐loaded samples. Elemental mapping provides additional confirmation of the homogeneous distribution of K, C, and N in both KPHI and 2%C₁N₁‐KPHI hybrid (Figure S5a,b). Quantitative elemental analysis reveals a C:N atomic ratio of approximately 1:1 in C₁N₁, and all composite samples show an increased carbon and nitrogen contents following C₁N₁ incorporation (Table S1). Together, these findings confirm the successful synthesis of C₁N₁ and its effective integration into a heterojunction structure with KPHI.

All composite samples exhibit increased specific surface areas compared to pristine KPHI, a result attributed to the inclusion of C₁N₁, which itself has a moderately higher surface area of 79 m² g⁻¹, as shown in Figure S6 and Table S2. This enhanced surface area is beneficial for photocatalytic processes, as it provides more active sites for reactant adsorption, thereby improving reaction kinetics and overall catalytic performance. To further examine the chemical structure and elemental composition, x‐ray photoelectron spectroscopy (XPS) was employed. The XPS survey spectra verified the presence of C, N, K, and O elements in all synthesized catalysts, with no significant elemental impurities detected (Figures S7a and S8a). For the pristine C₁N₁ material, the high‐resolution C 1s XPS spectrum revealed three distinct carbon environments with binding energies at 284.7 eV (C═C), 285.9 eV (N═C(C) ─N), and 287.2 eV (N─ (N)C═N) (Figure S7b). Furthermore, the N 1s spectrum was deconvoluted into three main peaks at 398.1 eV (N‐pyridinic), 399.7 eV (N‐quaternary), and 401 eV (─NH_x) (Figure S7c) [26]. To further elucidate the chemical structure of the hybrid C₁N₁‐KPHI catalysts, detailed high‐resolution XPS analyses were performed on all composite samples. The C 1s high‐resolution spectra for KPHI displayed characteristic peaks at 284.8 eV (C─C), 286.5 eV (C‐NH_x), and 288.16 eV (N─C═N) (Figure S8b) [27]. Across the composite series, a gradual decrease in the intensity of the 286.5 eV peak was observed, which correlates with a reduction in terminal ‐NH_x groups, as also supported by FTIR data. This attenuation does not stem from changes in cyano groups, whose spectral features remain unaltered. In the C₁N₁‐KPHI hybrids, a slight shift in the C 1s binding energies was detected relative to pristine KPHI, likely due to electronic interactions induced by the adjacent C₁N₁ phase. Notably, the N‐C = N peak shifted negatively by 0.09 eV in the 2%C₁N₁‐KPHI sample, suggesting an electron transfer between C₁N₁ and KPHI [28]. The N 1s high‐resolution spectra revealed four distinct components at 398.5 eV (sp²‐hybridized N in C─N═C), 399.8 eV (tertiary N, (N‐C₃), 400.9 eV (terminal ─NH_x or ─C≡N groups), and 403.5 eV (surface oxidized nitrogen) (Figure S8c) [29]. Additionally, a negative shift of 0.07 eV in the N═C─N peak of the N 1s spectra parallels the shift observed in the C 1s spectra, further reinforcing the evidence for electronic and chemical interactions between the two phases. This shift supports the occurrence of electron transfer from KPHI to C₁N₁, likely driven by the built‐in electric field at the heterojunction interface [30]. Furthermore, Table S3 reveals a decrease in the N₂C/N₃C ratio with increasing C₁N₁ content, suggesting the formation of nitrogen vacancies within the structure. To explore the role of these structural defects, electron paramagnetic resonance (EPR) spectroscopy was conducted on both pristine KPHI and the 2%C₁N₁‐KPHI hybrid (Figure S9). In both cases, the EPR spectra exhibited a primary Lorentzian signal with a g‐value of 2.003, characteristic of unpaired electrons localized on sp²‐hybridized nitrogen atoms in the π‐conjugated carbon nitride framework. The markedly stronger EPR signal in the 2%C₁N₁‐KPHI catalyst suggests a higher concentration of unpaired electrons [31]. This observation provides additional confirmation that structural defects are effectively introduced through hybridization with C₁N₁, thereby facilitating photoinduced charge transfer, which is advantageous for photoactivity.

2.2. Photocatalytic Performance

The photocatalytic performance of the synthesized materials for H₂O₂ generation was systematically evaluated under 410 nm LED irradiation, in the absence of a cocatalyst (Figure 2a). For evaluation, each catalyst (2.5 mg mL⁻¹) was suspended in an oxygen‐saturated aqueous glycerol solution (10% w/w). Among the tested samples, the 2%C₁N₁‐KPHI composite exhibited the highest activity, generating 32 mM of H₂O₂ after 1 h — a yield twice that of pristine unmodified KPHI (16 mM). Notably, glycerol oxidation produced glyceraldehyde (60 mM) as the major product, together with glycolic acid (20 mM), dihydroxyacetone (8.5 mM), and formic acid (5 mM) (Figure S10) [32, 33]. These results demonstrate that coupling photocatalytic H₂O₂ production with selective glycerol oxidation enables the simultaneous generation of value‐added oxygenates. A comprehensive investigation of the oxidation kinetics and the separation of the oxidation products from H₂O₂ will be pursued in future work.

FIGURE 2.

(a) Influence of C₁N₁ loading on the H₂O₂ evolution performance of KPHI. (b) Different reaction time of KPHI and 2%C₁N₁‐KPHI. (c) H₂O₂ evolution performance of C₁N₁, KPHI, C₁N₁ and KPHI (mixed by grinding), 2%C₁N₁‐KPHI. (d) Recyclability tests for 2%C₁N₁‐KPHI in H₂O₂ production. (e) Photocatalytic H₂O₂ evolution activities of KPHI and 2%C₁N₁‐KPHI using different excitation wavelengths, and (f) UV–vis spectrum and AQY values for 2%C₁N₁‐KPHI as a function of the irradiation wavelength.

Further increases in C₁N₁ content beyond 2 wt.% led to a gradual decline in photocatalytic performance, though still surpassing the activity of KPHI alone. This decline is attributed to excessive C₁N₁ loading, which causes local aggregation (as observed by SEM in Figure S2) and reduced crystallinity (broader and weaker XRD peaks in Figure 1b). These structural changes negatively affect charge transfer and reduce the intrinsic active sites of KPHI, ultimately resulting in decreased activity. Furthermore, a decline in H₂O₂ concentration after 1 h was observed for both pristine KPHI and 2%C₁N₁‐KPHI, which is attributed to the intrinsic H₂O₂ decomposition ability of the materials themselves under prolonged irradiation (Figure 2b). To further verify this, we performed H₂O₂ decomposition experiments under identical conditions. After 1 h of reaction, pristine KPHI decomposed 98.7% of the initially added H₂O₂, while the 2%C₁N₁‐KPHI composite decomposed only 57.6%, confirming that the composite effectively suppresses H₂O₂ decomposition (Figure S11). As a result, while the apparent activity of KPHI dropped to nearly zero after 100 min (i.e. the rates of H₂O₂ formation and decomposition reach a dynamic equilibrium), the 2%C₁N₁‐KPHI still retained considerable activity, producing 17 mM of H₂O₂. This further demonstrated that the epitaxially coupled C₁N₁‐KPHI catalyst effectively suppresses the undesired H₂O₂ decomposition pathway and maintains superior photocatalytic performance compared with KPHI during prolonged irradiation. To better understand the role of C₁N₁ in the composites, pristine C₁N₁ and a physical mixture of KPHI with 2%C₁N₁ (prepared by manual grinding, labeled as 2%C₁N₁‐KPHI‐M) were also tested (Figure 2c). Pristine C₁N₁ exhibited negligible photocatalytic activity, consistent with its intrinsic nature as a carbonaceous material lacking photocatalytic ability. For the physically mixed sample, UV–vis absorption spectrum in Figure S12 remained essentially unchanged compared with pristine KPHI, and the photocatalytic activity showed no significant enhancement. These results demonstrate that mere physical blending does not enable effective charge separation via directional electron transfer.

The improved performance of the 2%C₁N₁‐KPHI composite points to the formation of the built‐in electric CT field, which facilitated efficient electron transfer to O₂, driving the reduction reaction. 2%C₁N₁‐KPHI also demonstrated outstanding stability and recyclability. As shown in Figure 2d, the H₂O₂ concentration showed only a slight decrease over five consecutive reaction cycles, which we attribute to the minor catalyst loss during recovery between runs. Moreover, the main diffraction peaks of the recovered catalyst were well preserved in the XRD pattern (Figure S13), confirming the structural integrity of the material throughout the photocatalytic reaction. To evaluate the wavelength‐dependent photocatalytic behavior of 2%C₁N₁‐KPHI relative to KPHI, H₂O₂ generation tests were performed under various diode light sources (Figure 2e). The 2%C₁N₁‐KPHI catalyst exhibited a marked improvement in activity under simulated white light (λ ≥ 420 nm), delivering an activity 1.54 times higher than that of pristine KPHI. Under green LED illumination (λ = 528 nm), it also showed superior performance, reaching an activity 0.64 times higher than KPHI, with the less pronounced enhancement at longer wavelengths arising from the lower photon flux density of the green light source. These enhancements, particularly evident at longer wavelengths, suggest that the heterojunction between C₁N₁ and KPHI adds a further charge‐transfer band which effectively broadens the light absorption range of the photocatalyst. Furthermore, the apparent quantum yields (AQY) of 2%C₁N₁‐KPHI were determined to be 68.62% at 365 nm and 63.94% at 410 nm (Figure 2f), about two times higher than that of pristine KPHI (AQY_410nm = 30%). The performance achieved here outperforms most previously reported carbon nitride‐based photocatalysts (Table S4). However, we also want to point out that deviations from 100% AQY are mostly due to imperfect reactor engineering, as indeed stirred powder reactors show incomplete light absorption due to light transmission, light scattering and imperfect light focus [34, 35, 36], which make it very difficult to ensure that every photon is absorbed by the photocatalyst.

2.3. Optical and Electrochemical Characterization

To unravel the underlying reasons for the altered photocatalytic performance of the 2%C₁N₁‐KPHI hybrid, an extensive set of optical and photo(electro)chemical analyses was conducted. UV–vis diffuse reflectance spectroscopy (DRS) served as the primary method for initial optical assessment. As illustrated in Figure S14a, pure C₁N₁ exhibits broad spectral absorption, consistent with its black appearance and indicative of a well‐developed, graphite‐like π‐conjugated system [37]. In contrast, pristine KPHI shows pronounced absorption in the 300–450 nm range due to π → π ^* transitions, along with a secondary broad band in the 450–650 nm range, attributed to n → π ^* transitions enabled by K⁺ ions, which help reduce the molecular symmetry constraints [38]. Upon formation of the hybrid material, a visible color transition from yellow to yellowish‐brown is noted, accompanied by a red shift in the absorption edge that increases with higher C₁N₁ content (Figure 3a and Figure S14b). This shift is mainly ascribed to partial charge transfer involving the upper electrons of the PHI valence band, implying enhanced solar light absorption and a larger potential for photocarrier generation. To examine the charge separation and transfer processes in the 2%C₁N₁‐KPHI catalyst, a series of characterizations were conducted, including steady‐state photoluminescence (PL), time‐resolved PL (TRPL), electrochemical impedance spectroscopy (EIS) and transient photocurrent response (TPR). Steady‐state PL spectra reveal that pristine KPHI exhibits a strong emission peak, reflecting substantial radiative recombination of charge carriers (Figure 3b). Upon the addition of C₁N₁, the PL intensity notably declines, indicating suppressed charge recombination [39]. This phenomenon can be attributed to the electron transfer from KPHI to C₁N₁. TRPL analysis offers additional insight, showing a decrease in the average carrier lifetime from 1.47 ns for pristine KPHI to 1.13 ns for 2%C₁N₁‐KPHI (Figure 3c). This shorter lifetime reflects faster charge extraction, ultimately improving photocatalytic activity. As reveled by impedance analysis in Figure 3d, the smaller semicircle observed for C₁N₁‐modified PHI quantifies the lower charge transfer resistance compared with KPHI [40]. These improvements in charge separation and conductivity are further corroborated by TPR measurements (Figure 3e,f). The 2%C₁N₁‐KPHI composite demonstrates a significantly higher and more stable photocurrent than KPHI alone. Under 410 nm irradiation, its photocurrent is 2.5 times higher than that of pristine KPHI, and about 1.7 times higher under low‐intensity white light.

FIGURE 3.

(a) UV–vis DRS spectra of samples (inset: optical images of KPHI and x%C₁N₁‐KPHI powders). (b) Room temperature steady‐state PL emission spectra of all samples. (c) Solid‐state time‐resolved PL decay of KPHI and 2%C₁N₁‐KPHI. (d) Electrochemical impedance spectroscopy Nyquist plots of C₁N₁, KPHI and 2%C₁N₁‐KPHI. Transient photocurrent (e) λ = 410 nm and (f) white light for KPHI and 2%C₁N₁‐KPHI in 0.2 M Na₂SO₄ aqueous solution.

2.4. Mechanistic Investigation

To gain deeper insight into the dominant reaction mechanism within the synthesized composite, a series of reactive species quenching experiments were carried out. Furthermore, the reaction kinetics were systematically investigated by linear sweep voltammetry (LSV) using a rotating disk electrode (RDE), enabling detailed analysis of the electron transfer dynamics during photocatalytic H₂O₂ production. As shown in Figure S15, no H₂O₂ was detected either in the absence of light or without the catalyst, confirming the role of photocatalysis in the process. The requirement for oxygen was also clearly established: when the reaction system was purged with N₂, H₂O₂ production was minimal, while exposure to air resulted in substantial H₂O₂ production. In this system, H₂O₂ primarily proceeds through the reduction of molecular O₂ by photogenerated electrons. This process typically proceeds via two main pathways: a sequential two‐electron reduction (O₂ → ∙O₂ ⁻ → H₂O₂) or a direct two‐electron reduction (O₂ → H₂O₂) [41].

To identify the active intermediates involved in the reaction, radical scavenging experiments were conducted utilizing AgNO₃ (e⁻), tert‐butyl alcohol (TBA, ∙OH), and sodium thiosulfate (Na₂S₂O₃, ∙O₂ ⁻), as shown in Figure 4a [42, 43]. A marked decrease in H₂O₂ production upon addition of AgNO₃ highlighted the critical role of photogenerated electrons. TBA caused only a partial reduction in yield, indicating that ∙OH radicals are not the primary contributors to H₂O₂ generation. This conclusion is further supported by the continued detection of H₂O₂ in systems lacking glycerin or under nitrogen‐purged conditions. The presence of ∙OH radicals was nevertheless confirmed by EPR spectroscopy following light irradiation of the 2%C₁N₁‐KPHI catalyst (Figure S16), which implies that this channel exists as a side reaction. More importantly, the introduction of Na₂S₂O₃ led to an almost complete suppression of H₂O₂ formation, conclusively demonstrating that ∙O₂ ⁻ radicals are key intermediates in the reaction. These findings collectively suggest that the H₂O₂ generation process in the 2%C₁N₁‐KPHI system proceeds via two consecutive one‐electron reduction steps. This conclusion is directly supported by EPR spectroscopy, which showed no ∙O₂ ⁻ radicals in the dark, but a clear signal under 410 nm light exposure, with a significantly stronger response from 2%C₁N₁‐KPHI compared to pristine KPHI (Figure 4b). This indicates that the heterojunction facilitates more efficient O₂ to ∙O₂ ⁻ conversion, thereby boosting H₂O₂ production. To further probe the ORR kinetics and verify the proposed mechanism, RDE measurements were performed for both catalysts. In the same potential range, 2%C₁N₁‐KPHI exhibited higher cathodic current densities, signifying superior ORR kinetics (Figure 4c,d). Analysis using Koutecky–Levich (K‐L) plots revealed electron transfer numbers of 2.74 for KPHI and 2.03 for 2%C₁N₁‐KPHI. The value of 2.74 for pristine KPHI suggests that the oxygen reduction reaction proceeds via a mixed pathway, involving both a two‐electron reduction to H₂O₂ and a four‐electron reduction to H₂O, resulting in an apparent electron transfer number approaching 3. In contrast, the average value for 2%C₁N₁‐KPHI, being close to 2, strongly supports the predominance of a two‐electron reduction pathway leading to H₂O₂ formation, rather than the four‐electron pathway yielding water. The decrease in the K–L‐derived average electron transfer number from ∼3 for pristine KPHI to ∼2 for 2%C₁N₁‐KPHI reveals a significant mechanistic shift in the ORR. The construction of the dyadic structure suppresses the desorption of the ∙O₂ ⁻ intermediates, confining them to the surface and steering the reaction toward a 2e⁻ pathway for H₂O₂ production [44]. Based on all experimental data, we propose that the synergistic dyadic structure composed of KPHI (with a bandgap of approximately 2.7 eV) and C₁N₁ facilitates the formation of optically active charge‐transfer transitions (Figure 4e) [17, 18]. In this configuration, photoexcited electrons are preferentially transferred to charge‐transfer states predominantly localized on C₁N₁, while the corresponding holes remain electronically and spatially associated with the KPHI domains. This spatial charge separation significantly lowers the binding energy of excitons. As a result, a larger population of electrons can be efficiently trapped and stored in C₁N₁, thereby driving the two‐electron oxygen reduction toward H₂O₂.

FIGURE 4.

(a) Influence of different scavengers on the photocatalytic H₂O₂ production of 2%C₁N₁‐KPHI, (b) EPR spectra of DMPO‐∙O₂ ⁻ over KPHI and 2%C₁N₁‐KPHI, and LSV curves of (c) KPHI and (d) 2%C₁N₁‐KPHI tested by RDE (Inset: fitted K‐L plot). (e) proposed dyadic structure formed via charge‐transfer‐type orbital hybridizations at the KPHI and C₁N₁ interface.

To gain deeper insight into the interfacial charge transfer mechanism governing photocatalytic H₂O₂ evolution over the C₁N₁‐KPHI composite, density functional theory (DFT) calculations were performed. Initially, the structures of pristine KPHI, C₁N₁, and the composite C₁N₁‐KPHI were constructed and optimized, as shown in Figure 5a and Figure S17. Charge density difference analysis (Figure 5a and Figure S18) reveals significant charge redistribution at the interface upon composite formation, indicating strong electronic interaction between the two components. In Figure 5b, KPHI and C₁N₁ exhibit Fermi levels at −4.51 and −5.41 eV, respectively. The higher Fermi level of KPHI provides the thermodynamic driving force for electron transfer from KPHI to C₁N₁. Upon composite formation, the Fermi level shifts to −4.48 eV, which deviates from a simple average of the individual values, thus indicating that electrons flow from KPHI to C₁N₁ to establish a new electronic equilibrium. To further elucidate the electronic structure, density of states (DOS) and projected density of states (PDOS) analyses were conducted. Interestingly, DOS reveals that the electronic structure of the composite near the Fermi level closely resembles that of pristine KPHI rather than C₁N₁, indicating that KPHI serves as the primary contributor to the frontier electronic states. Meanwhile, PDOS further reveals strong orbital hybridization between the C‐p orbitals of C₁N₁ and the N‐p orbitals of KPHI at the interface (Figure 5b), providing a direct electronic coupling channel that facilitates interfacial charge transfer. In addition to these theoretical analyses, Mott–Schottky measurements were performed to investigate the band structure. As a carbonaceous material, C₁N₁ does not alter the intrinsic bandgap of KPHI, which is consistent with experimental observations in Figure 3a and Figure S19. Importantly, the band positions of the C₁N₁‐KPHI composite shift upward to a more negative potential compared with pristine KPHI in Figure 5c, further confirming the strong interfacial interaction and providing a thermodynamic advantage for the oxygen reduction reaction. To experimentally verify the reaction pathway, in‐situ FTIR spectroscopy was employed to monitor the intermediates during the photocatalytic process. The spectra clearly capture the characteristic signals of adsorbed O₂ (894 cm^{− 1}) and key intermediates, including ∙O₂ ⁻ (1108 cm^{− 1}), ∙OOH (1280 cm^{− 1}), and HOOH (1350 cm^{− 1}) [45, 46], confirming that the reaction proceeds via a superoxide‐mediated two‐electron reduction pathway from adsorbed O₂ to H₂O₂. Notably, the in‐situ FTIR results also show a dynamic reconstruction of interfacial water molecules under light irradiation, with a gradual transition from bound water (3270 and 2450 cm⁻¹) to free water (3600 cm⁻¹) over increasing illumination time [47, 48]. This interfacial water reconstruction could facilitate proton transfer during the ORR, thereby accelerating the catalytic kinetics for H₂O₂ generation [49, 50]. Collectively, these findings consistently support a selective two‐electron ORR pathway promoted by the interfacial charge‐transfer architecture.

FIGURE 5.

(a) DFT‐optimized structures for KPHI, C₁N₁, and the C₁N₁‐KPHI composite (with side‐view charge density difference on the right). (b) Projected density of states (PDOS) and corresponding Fermi levels of the as‐prepared samples. (c) Energy band positions of the samples derived from Mott–Schottky measurements. Time‐dependent in‐situ FTIR spectra of the 2%C₁N₁‐KPHI photocatalyst under light irradiation, showing (d) the evolution of signals from reactive intermediates and (e) the variation of surface‐adsorbed water signals.

3. Conclusion

In summary, this work presented a facile approach to improve the photocatalytic reactivity of ionic carbon nitrides through the rational construction of an epitaxial heterojunction. The designed composite effectively extended light absorption to the visible range and promoted exciton dissociation and electron transfer kinetics at the material surface, likely driven by a built‐in electric field. The resultant optimal catalyst, designated as 2%C₁N₁‐KPHI, exhibited excellent performance for light‐driven H₂O₂ generation as a model reaction, achieving an AQY of 64% at 410 nm without relying on any cocatalysts and with glycerol as a relatively sluggish hole scavenger. Mechanistic insights from optical and photoelectrochemical analyses confirmed that the heterojunction structure facilitated superior charge separation and transport, directing electrons preferentially toward the C₁N₁ component and mitigating charge recombination. Electrochemical characterization results further substantiated these observations by indicating accelerated ORR kinetics via a two‐electron process. Notably, DFT calculations confirmed the thermodynamic driving force for electron transfer and interfacial orbital hybridization, while in situ FTIR spectroscopy identified key intermediates (∙O₂ ⁻ and ∙OOH) and revealed interfacial water reconstruction that facilitates proton supply for ORR. More generally, this work underscores the significant potential of constructing heterojunctions based on ionic carbon nitrides as a promising pathway towards more active photocatalysts for a broad range of reactions.

Author Contributions

Chenyan Zhao: validation, formal analysis. Haijian Tong: conceptualization, methodology, writing – review and editing, writing – original draft, validation, investigation, formal analysis, data curation, visualization. Markus Antonietti: conceptualization, methodology, writing – review and editing, resources, project administration, funding acquisition, supervision. Christian Mark Pelicano: conceptualization, methodology, writing – review and editing, supervision, resources, project administration.

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting file: advs77182‐sup‐0001‐SuppMat.docx

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.