Self-Sustaining Ce(IV)/Ce(III) Cycle for Spontaneous Surface Hydroxyl Radical Generation Authors: Dexuan Li, Dahong Huang, Rongyue Wang, Jiapeng Zhong, Yuqiong Li, Xiaoyan Zhang, Yanan Wu, Yiyi Guo, Chuanhao Li, Jae-Hong Kim Published: 2026 ABSTRACT Advanced oxidation processes are indispensable for water remediation yet face a fundamental activity–selectivity trade-off, where reactive species are competitively quenched by complex aqueous matrices. While surface-confined chemistry, specifically utilizing surface-adsorbed hydroxyl radicals (•OHads), provides an appealing kinetic strategy to shield reactions from bulk-phase scavengers, its generation typically requires substantial energy inputs. Here, we report the spontaneous evolution of •OHads via a lattice-engineered 2D CeO2/βC3N4, comprising bilayer CeO2 and monolayer β carbon nitride (βC3N4). Unlike conventional CeO2, where surface Ce4+ sites are blocked by stable oxygen terminations (i.e., Ce4+O), the unique architecture of CeO2/βC3N4 stabilizes reactive Ce4+ sites via a surface superoxide layer (i.e., Ce4+-•O2 –). Upon dispersion into an aqueous system, this superoxide layer undergoes rapid protonation, exposing accessible Ce4+ sites for water oxidation to generate •OHads. Complementary in situ characterizations, isotope-labeled mass spectrometry, and theoretical calculations reveal thermodynamically favorable and stoichiometrically balanced one-electron oxygen reduction and water oxidation pathways, enabling self-sustaining Ce4+/Ce3+ redox cycling and continuous •OHads generation. This work establishes a pollutant-triggered paradigm for matrix-resilient water remediation by leveraging •OHads, eliminating the need for chemical additives or external energy inputs. Introduction The pervasive presence of recalcitrant pollutants in aqueous systems poses a substantial threat to ecological security and public health. Advanced oxidation processes (AOPs) are widely used for water remediation, conventionally employing aqueous hydroxyl radicals (•OHaq) for rapid decontamination. , However, the strong oxidative capacity of •OHaq, driven by its high redox potential (∼2.8 VRHE), also renders it susceptible to quenching by ubiquitous water-matrix constituents, such as chloride (Cl‑), bicarbonate (HCO3 –), and natural organic matter. To mitigate this limitation, recent studies have explored more selective oxidants such as singlet oxygen with a moderate redox potential (∼0.82 V). , While such oxidants are more tolerant to background scavengers, their lower oxidative strength can reduce effectiveness against electron-deficient contaminants. Developing reactive species that retain a high redox potential while resisting quenching by water matrices therefore remains an ongoing need. In this context, an alternative strategy has emerged that leverages surface-bound high-valent metal species with strong oxidizing capacity (e.g., E Cr(VI)‑oxo (1.51 VRHE), E Mn(VII)‑oxo (2.05 VRHE) and E Fe(VI)‑oxo (2.20 VRHE)). , This strategy can suppress matrix interference by physically confining the reaction to the catalyst interface, thereby reducing exposure of reactive species to bulk-phase scavengers. , However, such frameworks still require continuous addition of peroxide precursors, raising concerns about cost, safety, and sustainability. Fluctuating wastewater profiles and the demand for complete decontamination may necessitate peroxide overdosing, which can increase secondary environmental risks. Moreover, the reliance on these reagents entails additional hazards associated with peroxide storage and transport, as well as carbon emissions from centralized production and distribution. Surface-adsorbed hydroxyl radicals (•OHads) have been proposed as a theoretical optimum that integrates a high oxidative potential (∼1.5 VRHE) with improved matrix tolerance enabled by a surface-confined strategy. More critically, •OHads evolution can be achieved using external energy inputs (e.g., light or electricity) via the oxidation of abundant OH–/H2O rather than peroxides. − The chemical-free nature of these approaches eliminates the potential risks associated with the use of precursor chemicals, and the rapid operational response can enable controllable generation of •OHads in response to real-time changes in water quality. Nevertheless, the scalable implementation of these technologies still faces practical limitations. Photocatalysis often suffers from limited light delivery and modest overall process efficiency. Electrocatalysis is generally constrained by low water conductivity and is prone to electrode deactivation. Notably, both approaches share a common challenge of relatively high energy consumption, which remains a key barrier to their broader industrial adoption. Achieving efficient •OHads generation without supplying chemical reagents or external energy is a highly espoused but difficult-to-attain goal. Thermodynamically, specific metal redox couples (e.g., Co3+/Co2+ (1.92 VRHE) and Ce4+/Ce3+ (1.75 VRHE)) possess sufficient potential to oxidize H2O into •OHads. Among candidate materials, cerium oxide (CeO2) is particularly attractive for its crustal abundance and environmental benignity. ,− Nevertheless, in conventional CeO2, reactive Ce4+ sites are typically buried within the bulk lattice or passivated by stable surface terminations (i.e., Ce4+=O). The coordinative saturation can render Ce4+ less active toward •OHads production. Therefore, material engineering to expose accessible Ce4+ may provide a route to unlocking this latent reactivity. Here we introduce a two-dimensional (2D) CeO2/βC3N4 catalyst consisting of bilayer CeO2 and monolayer β carbon nitride (βC3N4), in which the unique architecture creates a superoxide layer that undergoes rapid protonation upon aqueous contact and thus unlocks reactive Ce4+ sites. We evaluate catalytic performance for the degradation of multiple environmental contaminants across complex water matrices. Through a combination of in situ electron paramagnetic resonance (EPR), operando Raman spectroscopy, and isotope-labeled mass spectrometry, we elucidate the mechanistic basis for continuous generation of •OHads without chemical additives or external energy. This work demonstrates a self-sustaining strategy driven exclusively by pollutants as the ultimate electron donor, providing a transformative framework for next-generation water purification without resource-intensive inputs. Experimental Method Catalyst Preparation The gC3N4 was prepared by the thermal polymerization method. Melamine (5 g) was transferred into a mortar and ground with a pestle until the mixture appeared to be a homogenous white powder. This powder was then loaded into a crucible and properly sealed. The reaction vessel was placed vertically into a muffle furnace and heated at a rate of 5 °C/min to 550 °C for 4 h. The furnace was then shut off and allowed to radiatively cool to room temperature. The crude product was repeatedly washed with deionized water. The pure gC3N4 was obtained as a homogenous light-yellow powder. The 2D CeO2/βC3N4 was prepared by the molten salt method. CeCl3·7H2O (0.89 g), gC3N4 (0.6 g), KCl (3.3 g) and LiCl (2.7 g) were transferred into a mortar and ground with a pestle until the mixture appeared to be a homogenous light-yellow powder. This powder was then loaded into a crucible. The reaction vessel was placed vertically into a muffle furnace and heated at a rate of 5 °C/min to 550 °C for 4 h. The furnace was then shut off and allowed to radiatively cool to room temperature. After cooling to room temperature, the product was washed with water for 12 h at 80 °C and dried at 60 °C overnight. The pure 2D CeO2/βC3N4 was obtained as a yellow powder. Results and Discussion Material Synthesis and Characterization We loaded CeO2 onto C3N4 to obtain CeO2/βC3N4 composite catalyst using a one-step molten-salt method. , Briefly, gC3N4 was mixed with cerium chloride (CeCl3) along with auxiliary salts such as lithium chloride (LiCl) and potassium chloride (KCl), which decreased the melting point of the salt mixture (Figure A). As the temperature increased to 350 °C, the molten salts enabled the uniform distribution of Ce3+ ions (from the CeCl3 precursor) into the interlayer of gC3N4. Upon reaching 450 °C, Ce3+ ions confined within the gC3N4 interlayers underwent oxidation by O2 to Ce4+ (in the CeO2 form), accompanied by the simultaneous exfoliation of multilayer gC3N4. When the temperature further increased to 550 °C, gC3N4 recrystallized into βC3N4, which was confirmed by the emergence of a series of XRD peaks at 16.0, 27.8, 32.2, 43.1, and 45.2° which correspond to the (100), (2–10), (200), (3–10) and (300) planes (Figure B). Notably, all other signature peaks along the z-axis (i.e., the (101) plane at 40.9° and the (2–11) plane at 47.2°) disappeared, suggesting the formation of monolayer βC3N4 confined to the xy-plane (detailed lattice parameters are presented in Table S1). The selected area electron diffraction (SAED) pattern (Figure S1) exhibited concentric diffraction rings, consistent with the characteristic planes observed in XRD diffractograms for βC3N4 and CeO2. (A) Schematic illustration of the synthetic strategy for CeO2/βC3N4. (B) X-ray diffraction (XRD) patterns for CeO2, gC3N4 and CeO2/βC3N4. The detailed lattice parameters are presented in Table S1. (C) Structural models of untwisted and twisted CeO2 on βC3N4 support (defined by the aligned x- and y-axes of the lattice box), accompanied by the corresponding high-resolution transmission electron microscopy (HRTEM) images for each configuration. (D) The height profile and the stereoscopic morphology of CeO2/βC3N4 characterized by atomic force microscopy (AFM). Transmission electron microscopy (TEM) observations (Figure S2), combined with energy-dispersive spectroscopy (EDS) mapping (Figure S3), indicated the uniform distribution of CeO2 nanosheets on the gauze-like βC3N4 substrate. Moreover, this nanosheet morphology represents a distinctive structural feature of CeO2/βC3N4, rather than an incidental structure present in the pristine CeO2 or gC3N4 components (Figure S2). Certain CeO2 nanosheets exhibited darker contrast compared to others, suggesting variations in layer thickness (Figure S4). HRTEM, combined with theoretical modeling, provided further insight into the CeO2/βC3N4 structure at the atomic scale. As illustrated in Figure C (top left panel), the twisted alignment between pristine CeO2 (220) and βC3N4 (100) unit cells along the x- and y-axes , led to a substantial lattice mismatch of 24 and 15% along the x- and y-axes, respectively (Figure S5). In contrast, the 36°-twisted configuration between CeO2 (220) and βC3N4 (100) planes showed closer lattice matching, with the degree of lattice mismatching reduced to 4 and 3% along the x- and y-axes (Figure C, top left panel), respectively. This optimized interfacial alignment is expected to facilitate selective in-plane growth while effectively suppressing out-of-plane crystallization along the z-axis, leading to the formation of ultrathin CeO2. The multilayer architecture was confirmed by the presence of Ce atoms from vertically stacked layers (Figure C, bottom left), while the bilayer structure was further evidenced by the correspondence between observed Ce positions and theoretically predicted locations (Figure C, bottom right). Atomic force microscopy (AFM) analysis of the CeO2/βC3N4 composite (Figure D), focusing on the predominant ultrathin CeO2 nanosheets as well as the occurrence of sporadic thick layers, further revealed a well-defined hybrid layer structure with thicknesses of 0.54 and 0.53 nm. Considering the theoretical thickness of monolayer (0.47 nm) and bilayer (0.94 nm) βC3N4, and bilayer CeO2 (0.56 nm), we propose that CeO2/βC3N4 primarily consists of bilayer CeO2 and monolayer βC3N4. , Multiple locations were analyzed to collect statistically representative data, as shown in Figure S6. FT-IR spectroscopy analysis revealed variations in characteristic chemical bonding between βC3N4 and CeO2 components. As shown in Figure A (detailed peak information in Table S2), the C–O stretching at 989 cm–1 in gC3N4 shifted to 1070 cm–1 in CeO2/βC3N4. , This blue shift indicated a significant increase in electron density, corroborating the occurrence of charge transfer from βC3N4 to CeO2 through the C–O–Ce4+ linkage. Such electron redistribution was further evidenced by the downfield peak shifts (i.e., CI, 156.7 to 168.8 ppm; CII, 163.3 to 174.9 ppm) in solid-state 13C nuclear magnetic resonance (NMR). Specifically, the decreased electron density around carbon atoms likely originated from the interfacial electron transfer from βC3N4 to CeO2 (Figure S7), contributing to the low impedance of CeO2/βC3N4 composite (Figure S8). The surface CeO2 layer also induced the conversion of hybrid CN moieties (e.g., C–N/CN, –C≡N, –CNH, etc.) into a uniform CN framework of four-coordinate carbon bonded to N atoms, as shown by the transformation of multiple peaks ranging from 1100 to 1320 cm–1 in gC3N4 into a broad peak in CeO2/βC3N4. The emergence of a Ce4+–O–C peak (288.4 eV) and disappearance of –NH species (401.1 eV) were further validated through XPS (Figure S9 and Table S3). (A) Fourier transform infrared (FT-IR) spectra of CeO2, gC3N4 and CeO2/βC3N4. (B) Solid-state electron paramagnetic resonance (EPR) spectra of CeO2, gC3N4 and CeO2/βC3N4 and the corresponding simulated spectra obtained using EasySpin toolbox in MATLAB (•O2 –: g x = 2.0052, g y = 2.0062, g z = 2.0500, g x strain = 0.008, g y strain = 0.003, g z strain = 0.015, line width = 1.5; NO•: g x ≈ g y = 1.9994, g z = 2.0029, AN 14 = [0 0 50.4], AStrain = [0 0 30], line width = 0.8; unpaired electrons: g = 2, line width = 1). (C) Raman spectra of CeO2 and CeO2/βC3N4 (light source wavelength: 785 nm, the potential interference from gC3N4 was ruled out by control experiments (Figure S9E)). (D) O 1s X-ray photoelectron spectroscopy (XPS) spectra of CeO2 and CeO2/βC3N4. Solid-state EPR was subsequently conducted to understand the influence of electronic interaction on the local environment of CeO2/βC3N4 (Figure B). The abundant unpaired π-conjugated electrons in gC3N4 led to a prominent peak at g = 2, which decreased upon binding with CeO2. The accumulated electrons on CeO2 resulted in the formation of exposed Ce3+, which underwent rapid oxidation by O2 to form surficial superoxo species (i.e., Ce4+–•O2 –), as evidenced by a series of signals with characteristic g-values (g x = 2.0052, g y = 2.0062, g z = 2.0500). Previous studies suggest that the remaining signals can be attributed to nitroxide (NO•) , formed through electron transfer from nitrogen in βC3N4 to CeO2 via Ce4+–O–N bonds, which subsequently activates O2 to generate Ce4+–•O2 –. The peak at 463.3 cm–1 in Raman spectra (Figure C and Table S4) corresponded to the triply degenerate F2g mode of CeO2, a symmetric breathing mode involving the eight coordinating oxygen atoms. In conventional CeO2, oxygen atoms originated solely from lattice oxygen (i.e., Ce4+–O(8)). In contrast, in CeO2/βC3N4, oxygen on the top layer of Ce4+ consisted of six lattice oxygen and two oxygen atoms from surface •O2 – (i.e., O(6)–Ce4+–•O2 –). However, the presence of βC3N4 beneath the CeO2 is expected to prevent Ce4+ in the bottom layer from forming Ce4+–•O2 – moieties, leading to a deviation from the centrosymmetric position of eight coordinating oxygen environment. This structural distortion was evidenced by the DII peak at 595.5 cm–1. Another characteristic peak at 800.4 cm–1 suggested the high O2 affinity of CeO2/βC3N4, facilitating the subsequent conversion into •O2 –. Quantitative analysis of the O 1s XPS spectra (Figure D) confirmed the differences in oxygen species between conventional CeO2 and CeO2/βC3N4. Conventional CeO2 exhibited two primary peaks at 528.7 and 530.9 eV, , corresponding to lattice oxygen (68.6%) and Ce3+–OH– (H2O) (31.4%), respectively. In CeO2/βC3N4, additional peaks assigned to Ce4+–O–C and Ce4+–•O2 – were detected at 531.2 eV and 532.1 eV, respectively. Notably, although the overall Ce3+/Ce4+ ratio remained comparable between the two materials (Figure S9 and Table S3), a high fraction of Ce4+ (21.6%) existed as reactive Ce4+–•O2 – species in CeO2/βC3N4, rather than inert lattice Ce4+ in CeO2. Pollutant Degradation As shown in Figure A, CeO2 and gC3N4 did not measurably degrade BPA, whereas CeO2/βC3N4 achieved nearly complete BPA removal (exceeding 90%) within 1 h. Moreover, CeO2/βC3N4 retained a comparable activity across a broad pH range from 3 to 11 (Figure S10), highlighting its robustness under diverse environmental conditions. Control experiments ruled out contributions from adsorptive BPA removal, direct stoichiometric oxidation by Ce4+, H2O2 and light-driven reactions (Figures S11–S12), and the negligible change in kinetics upon replacing pure O2 with air (Figure S12) suggested that O2 was not a limiting factor. (A) Comparative degradation performance of BPA using CeO2, gC3N4 and CeO2/βC3N4. (B) EPR spin-trapping analysis in the CeO2/βC3N4 system. 2,2,6,6-tetramethylpiperidine (TEMP) was used to trap singlet oxygen (1O2), forming the 2,2,6,6-tetramethylpiperidin-1-oxyl (TEMP-O) adduct. 5,5-dimethyl-1-pyrroline N-oxide (DMPO) served as the spin trap for •OH and •O2 –, yielding the DMPO–OH and DMPO-O2 – adducts, respectively. Experiments were conducted in aqueous solution or pure DMPO to distinguish solvent-dependent •O2 – generation. (C) BPA degradation performance with different quenchers. (D) Degradation performance for various pollutants using CeO2/βC3N4. (E) BPA degradation performance with the addition of different concentrations of NaNO3 (1 mg/L), NaHCO3 (40 mg/L), NaCl (1 mg/L) and FA (1 mg/L). Experimental conditions: 0.5 g L–1 catalyst, pH = 7, room temperature (about 20 °C), pure O2 flow rate = 100 mL/min, atmospheric pressure in an open system. Error bars represent standard deviations from triplicate experiments. Results from recycling experiments supported catalyst longevity, a critical requirement for translation to practical application. CeO2/βC3N4 retained a high activity up to 97% over extended use for up to 40 cycles (Figure S13A). Moreover, CeO2/βC3N4 preserved ∼95% activity after three months of storage, reaffirming the stability essential for practical applications (Figure S13B). Marginal Ce leaching indicated excellent structural stability (Figure S14). In addition, unlike many oxidative catalysts that suffer from gradual oxidation, negligible variation in the Ce oxidation state was detected across cycles (Figure S15 and Table S5). This behavior was consistent with the role of Ce in mediating electron transfer; the abrupt current decrease upon O2 introduction during open-circuit current analysis corresponded to ORR, whereas the sustained current increase following BPA addition was attributable to BPA oxidation (Figure S16), providing qualitative evidence of how electron balance was maintained in the CeO2/βC3N4 system. We further employed EPR analysis to distinguish the roles of different reactive oxygen species (ROS) in BPA oxidation. As shown in Figure B, the absence of the TEMP-O signal excluded the involvement of 1O2, while the appearance of the DMPO–OH signal indicated •OH formation. Using anhydrous DMPO to minimize water interference in •O2 – detection, we observed the DMPO-O2 – adduct, which suggested the presence of •O2 – in the system (Figure S17). The consistent signal intensity of Ce4+-•O2 – before and after reaction supported that CeO2/βC3N4 sustained the aforementioned surface motif (Figure S18). We note that the generation of •OHaq requires a high redox potential (∼2.8 VRHE ), which cannot be solely driven by the Ce4+/Ce3+ redox couple (1.75 VRHE ). In contrast, the formation of surface-adsorbed •OH (•OHads) has been reported to occur at a significantly lower potential (∼1.5 VRHE , ), which is thermodynamically accessible relative to the Ce4+/Ce3+ redox potential − via water oxidation (WOR). Therefore, we examined which radical species among •O2 –, •OHaq or •OHads contributes to BPA degradation, by introducing various scavengers into the reaction system: BQ (p-benzoquinone, •O2 – quencher), isopropanol (IPA, •OHaq quencher) and potassium iodide (KI, •OHads quencher) (Figure C). The minor inhibition in 1-hour degradation efficiency upon IPA addition (∼10%) excluded a dominant role of •OHaq. In contrast, KI led to a marked decrease in degradation efficiency (∼92%), indicating a primary contribution from •OHads. The proposed •OHads-driven pathway for BPA oxidation aligned well with the reaction intermediates identified by high-performance liquid chromatography mass spectrometry (HPLC-MS) analysis (Figures S19–S20 and Table S6). Interestingly, BQ caused an even greater inhibition (up to ∼97%), which can be attributed to BQ-mediated poisoning of Ce4+ sites, either via the CO groups in BQ or semiquinone radicals generated from its reaction with •O2 –. This interference eliminated both •OHads-driven decontamination and the additional, albeit slower, contribution of •O2 – to BPA degradation. Furthermore, CeO2/βC3N4 demonstrated broad applicability by effectively degrading a suite of representative pollutants (Figure D), highlighting its potential as a versatile catalyst for environmental remediation. To further evaluate the practical applicability, degradation experiments were conducted in the presence of common background anions and NOM typically found in natural water bodies (Figures E and S21). In contrast to conventional Fenton systems where the presence of Cl‑, HCO3 –, NO3 –, or fulvic acid (FA) can significantly inhibit degradation performance (even exceeding 60% inhibition), , the •OHads-mediated system here maintained >95% removal efficiency, highlighting the matrix tolerance of •OHads. These results suggest that confining •OH radicals to the material surface as adsorbed •OHads can effectively prevent quenching by aqueous matrix components, thereby enhancing selectivity toward target pollutants and strengthening the catalyst’s potential for practical environmental applications. Mechanistic Insights into ORR and WOR We employed in situ EPR spectroscopy to examine the generation and transformation of •O2 – and •OHads on CeO2/βC3N4 during the reaction. In an anhydrous DMPO system (top panel, Figure A), which allowed only ORR to occur and prevented WOR, the DMPO-O2 – signal first decayed within 5 min under anaerobic conditions and then reappeared after O2 purging for another 5 min. The DMPO-O2 – signal was fully restored after oven drying, mirroring the treatment applied during recycling intervals and aligning with the aforementioned catalytic longevity (Figure S13). (A) In situ EPR spectra of the CeO2/βC3N4 system in DMPO and water. DMPO-O and DMPO-O2 – originate from the reaction between DMPO and •O2 – under aqueous and anhydrous conditions, respectively. (B) In situ Raman spectra of CeO2/βC3N4 during the reaction (The selected regions of bilayer CeO2 structures were characterized to evaluate their catalytic performance.). (C) In situ FT-IR spectra of O2 activation via CeO2/βC3N4. (D) The proportion of •OHads sources inferred from the isotopic distribution of coumarin oxidation products, as determined by HPLC-MS. Inset: HPLC-MS spectrum of 7-hydroxycoumarin generated from the reaction between coumarin and •OHads. (E) Open-circuit potential curves (OCP) on CeO2/βC3N4 electrode (Electrolyte: 0.05 M Na2SO4 aqueous solution). Interestingly, in water, 5 min of N2 purging eliminated any detectable signal (bottom panel, Figure A). We attribute this observation to the inherent instability of •O2 – in aqueous solution and the suppressed •O2 – generation under O2-free conditions. Subsequent O2 purging for 30 seconds rapidly generated a series of overlapping peaks, which were deconvoluted into DMPO-O and DMPO–OH through signal fitting, corresponding to the coexistence of •O2 – and •OHads (Figure S22) , at a ratio of 1:2.3. Distinct from DMPO-O, which exclusively originated from •O2 – generated during ORR, DMPO–OH could form through two pathways: (i) direct generation through trapping of •OHads generated from one-electron WOR, and (ii) indirect formation from one-electron ORR (DMPO + •O2 – → DMPO-OOH → DMPO–OH). To differentiate these pathways, we performed experiments in dimethyl sulfoxide (DMSO), a selective •OH quencher, to suppress the first pathway (Figure S23). The results demonstrated that the ratio of DMPO-O and the accompanying DMPO–OH (both derived from the DMPO-OOH conversion) was approximately 1:1. This result allowed us to determine the actual ratio of ORR to WOR as 1:(2.3-1), representing 43.2% ORR and 56.8% WOR. EPR results under various pHs (Figure S24) further substantiated the efficient generation of •OHads across a broad pH range, consistent with the pH-independent catalytic activity of CeO2/βC3N4 observed in Figure S10. Upon addition of BPA to the test system, the DMPO–OH signal disappeared, indicating that •OHads rapidly reacted with BPA to abstract electrons, thereby replenishing the electron pool and sustaining the catalytic cycle. As shown by in situ Raman analysis (Figure B), the characteristic peak of O(6)-Ce4+-•O2 – at 463.3 cm–1 rapidly disappeared upon the introduction of BPA, H2O and N2, indicating the depletion of Ce4+-•O2 – during BPA oxidation and a consequent rearrangement of local structure into asymmetric Ce3+-O(6). Continuous O2 purging for another 10 min resulted in the reemergence of the O(6)-Ce4+-•O2 – peak, supporting the reversible transformation of Ce4+-•O2 –. Similarly, the peak at 373.2 cm–1, attributed to the Ce4+-•O2 – moiety, underwent reversible decay and regeneration. The dynamic equilibrium of Ce4+-•O2 – during the reaction was consistent with the XPS results, which indicated minimal changes in the Ce4+-•O2 – species content before and after reaction (Figure S15 and Table S5). Time-resolved FT-IR spectroscopy (Figure C) further provided spectroscopic evidence for the generation of Ce3+-•OHads from Ce4+–OH–/H2O. Specifically, two characteristic bands at 3701.2/3727.0 cm–1 were detected upon exposure to O2/H2O vapor, corresponding to Ce4+–OH– stretching modes. The peaks at 3602.6/3626.2 cm–1 were assigned to Ce3+-•OHads. The addition of BPA/O2/H2O to the reaction system resulted in the disappearance of the Ce3+-•OHads signal, demonstrating that •OHads generated from WOR can react directly with BPA (Figure S25, statistics in Table S7). This indicates that although •OHads is surface-confined, it can extract electrons from BPA via interfacial electron transfer. Consequently, the catalyst ultimately harvests electrons from the pollutant to sustain its redox cycle, consistent with the analysis in Figure S16. The above results collectively suggested that •O2 – and •OHads were generated from the one-electron ORR and WOR, respectively. We verified their occurrence at reactive Ce sites through the 18O-labeled Raman spectroscopy (Figure S26), followed by quantifying the relative contributions of the two routes by isotope-labeled HPLC-MS (Figures D and S27). Using 10% H2 18O as the solvent and coumarin as a •OHads probe, 5.43% of •OHads adducts incorporated one 18O atom. Given the 10% isotopic abundance, this corresponded to 54.3% of •OHads originating from WOR, with the remaining 45.7% originating from surface •O2 – transformation. The near-stoichiometric 1:1 ratio (54.3 vs 45.7%) between the two pathways was mechanistically critical, as it maintains electron balance between one-electron ORR (•O2 – generation) and WOR (•OHads production), thereby preserving the oxidation state of the catalyst. Notably, this quantitative evidence additionally confirmed that •O2 – converted to •OHads via direct O–O bond cleavage (eq ) , rather than the conventional 2-electron pathway (eqs –). Any additional electron transfer beyond the coupled ORR/WOR processes would otherwise disrupt this equilibrium, potentially leading to progressive CeO2/βC3N4 oxidation and consequent loss of catalytic activity. 1 Ce4+‐O2−•+H+→Ce4++OHads•+Oads 2 O2−•+e−+2H+→H2O2 3 H2O2+e−→OH•+OH− Real-time open-circuit potential monitoring (OCP, Figure E) revealed three distinct regimes: (i) an O2-induced potential increase resulting from the oxidation of exposed Ce3+ into Ce4+, (ii) a NaOH-driven potential decrease due to the enhanced WOR via H+ consumption (eq ), and (iii) H2SO4-mediated potential drop attributed to WOR inhibition through H+ accumulation. These observations were consistent with the coupled ORR-WOR redox cycling within the designed catalytic system. Specifically, the first step involves the Ce3+-driven reduction of adsorbed O2 to form a Ce4+-•O2 – moiety. Subsequently, the exposed Ce4+ oxidizes OH– to generate •OHads, simultaneously regenerating Ce3+ to sustain the redox cycle of Ce3+/Ce4+ (Figure S28). Free Energy Changes along ORR and WOR We performed DFT calculations to examine the adsorption of O2 and H2O, which are the critical initial steps for ORR and WOR, respectively. In situ XRD analyses upon BPA and O2 introduction (Figure S29) identified the reactive (220) plane, providing critical structural insights for constructing the most relevant model for density functional theory (DFT) calculations (the optimized structure in Table S8). On pristine CeO2 (Table S9), the adsorption energy (ΔE ads) for H2O is –1.52 eV, much more favorable than that of O2 (+1.57 eV), indicating a thermodynamic preference for H2O adsorption (Table S10). Moreover, the two O–H bonds in the adsorbed H2O remain equivalent at 0.98 Å and H–O–H bond angle of about 106.5°, suggesting that the H2O molecule does not dissociate upon adsorption. These simulations suggest that CeO2 is likely to be passivated by a stable surface H2O layer, consistent with the previous studies. In contrast, on CeO2/βC3N4 (Table S11), adsorbed H2O displayed bond asymmetry (1.06 and 0.97 Å for the –H and −OH fragments, respectively, along with an expanded H–O–H angle of about 111.6°), indicating efficient H2O activation. Although H2O shows a more favorable adsorption energy (ΔE ads = –1.92 eV) than O2 (ΔE ads = –1.21 eV), the rapid activation of H2O may mitigate surface poisoning of CeO2/βC3N4. The negative ΔE ads for O2 indicates a thermodynamically favorable, exothermic adsorption process, serving as a prerequisite for the subsequent O2 activation. These findings highlight that CeO2/βC3N4 promotes both H2O activation and O2 adsorption, endowing it with dual catalytic functionality for both ORR and WOR. Gibbs free energy calculations further confirmed the thermodynamic spontaneity of both one-electron ORR and WOR pathways on CeO2/βC3N4 (Table S12). The ORR mechanism initiates with an exergonic (1.30 eV) electron transfer from Ce3+ to adsorbed O2, forming a Ce4+-•O2 – intermediate, followed by protonation to yield Ce4+-•OOH (ΔE = –0.99 eV, Figure A) and subsequent O–O bond cleavage that generates •OHads (Figure S30 and Table S13). The exposed Ce4+ active sites after ORR facilitate an energetically favorable conversion of one-electron WOR (ΔE = –0.85 eV), producing an additional •OHads (Figure B). Notably, the adsorption energies in Table S10 differ slightly from those in Figure A–B, due to the additional inclusion of zero-point energy and thermal corrections that account for the vibrational entropy of adsorbed species during the pathway analysis. Reaction free energy diagrams for the one-electron (A) ORR and (B) WOR process on the surface of CeO2/βC3N4. (C) Adsorption energies of H2O and O2 on CeO2 and CeO2/βC3N4 and the schematic illustration of the mechanism for one-electron ORR and WOR driven by the Ce3+/Ce4+ redox cycle. Together, these computational results are consistent with the experimental observation of self-sustaining ORR-WOR cycling and Ce3+-Ce4+ redox activity under ambient conditions, establishing CeO2/βC3N4 as an effective bifunctional catalyst for autonomous redox cycling (Figure C). In a typical run, one of the lone-pair electrons from N in βC3N4 is transferred to CeO2, generating surface Ce3+. This Ce3+ species is rapidly oxidized by O2 from air to form surface superoxo. Given the intrinsic instability of •O2 – in aqueous systems, it undergoes sequential reactions (eqs –), reopening the Ce4+ state, which subsequently oxidizes water to yield Ce3+-•OHads. Following the reaction of contaminants with •OHads, Ce3+ becomes accessible for the next cycle. This work presents a self-sustaining approach that circumvents the fundamental trade-off between oxidative power and matrix compatibility in AOPs, enabling efficient pollutant degradation without requiring chemical additives or external energy inputs. The 2D CeO2/βC3N4 catalyst, comprising bilayer CeO2 precisely aligned on monolayer βC3N4 through twisted lattice engineering, enables autonomous and concurrent one-electron ORR and WOR reactions to form •OHads. The unique architecture leverages unpaired electrons from βC3N4 to reduce Ce4+ into Ce3+, driving ORR to produce •O2 –. Subsequent •O2 – release regenerated highly oxidative Ce4+ sites, facilitating rapid WOR to generate •OHads. This stoichiometrically balanced electron transfer maintains an autonomous Ce3+/Ce4+ redox cycle without net consumption of reduced catalyst sites, which is fundamentally different from most transition metal catalysts used in similar reaction schemes. This work establishes a pollutant-triggered paradigm for matrix-resilient water remediation that obviates reliance on chemical additives and external energy inputs, addressing critical barriers to sustainable environmental remediation. Environmental Implication The practical application of advanced oxidation processes for water remediation is constrained by a fundamental trade-off between the high reactivity of hydroxyl radicals and their susceptibility to quenching by ubiquitous water matrix constituents. This work introduces a lattice-engineered 2D CeO2/βC3N4 catalyst that spontaneously generates surface-adsorbed hydroxyl radicals (•OHads) through coupled one-electron oxygen reduction and water oxidation reactions. The •OHads achieve high oxidative power while being effectively shielded from bulk-phase scavengers, maintaining >95% degradation efficiency in complex water matrices. Critically, this self-sustaining system operates without chemical additives or external energy, using pollutants as the ultimate electron donor. By eliminating reliance on peroxides and external energy inputs, this work addresses key barrierschemical safety hazards, carbon emissions, and energy coststhat currently limit scalable, decentralized water treatment. Supplementary Material