Dual-site metal-free catalyst for generation and utilization of 1O2
Given the notable oxygen-binding affinity and ease of formation of NVs, we chose NVs as reaction sites to design the metal-free dual-site catalyst for 1O2 generation. Motivated by fluorine’s exceptional ability to modulate electron behavior25, we strategically positioned F near diverse NVs within the carbon matrix, referred to as Nv-NFC, contrasting with the baseline Nv-NC (Fig. 1a). During theoretical optimizations, the F atoms shifted slightly, yet the overall structure remained stable (Supplementary Fig. 1). DFT calculations showed significant modulation in the electronic configuration on the fluorinated catalyst’s surface (Fig. 1b and Supplementary Fig. 2). F’s high electronegativity and potent electron affinity led to localized charge distributions, creating distinct electron-rich and electron-deficient zones, disrupting the symmetry of surface electron distribution. This asymmetry would further benefit the adsorption of reactants and intermediates, thus contributing to the optimization of catalytic performance26,27,28,29. Moreover, various NV types displayed unique electronic structures (Fig. 1b and Supplementary Fig. 2), illustrating that precise electronic tuning at the active site is achievable through F and N synergy.Fig. 1: Theoretical activity of Nv-NFC.a Schematic of nitrogen vacancy and fluorine doping. b The electrostatic potential distributions for Nv-NC-2 (left) and Nv-NFC-2 (right). c Schematic representation for PMS activation, using the Nv-NFC-2 structure as an example. d, e Free energy diagram for PMS activation on different nitrogen vacancies without (d) and with (e) F doping. f Difference charge density for PMS adsorption on Nv-NC-2 (left) and Nv-NFC-2 (right). Yellow and cyan regions represent electron accumulation and electron depletion, respectively. Grey, blue, cyan, red, yellow, and white spheres represent C, N, F, O, S, and H atoms, respectively. g Free energies for PMS adsorption and phenol adsorption on different nitrogen vacancies with and without F doping. Source data are provided as a Source Data file.This synergistic electronic adjustment significantly differentiated the catalytic structures’ capabilities to adsorb and activate PMS. A schematic in Fig. 1c shows the optimal configurations for PMS activation, where the PMS adsorption energies for the three F-free catalysts were positive (Fig. 1d; 0.48, 0.93, and 0.81 eV, respectively), indicating ineffective PMS adsorption. In contrast, fluorinated catalysts showed negative adsorption energies (−3.75, −1.01, and −0.58 eV, respectively) (Fig. 1e), suggesting that introducing F to induce asymmetry is an effective strategy to enhance adsorption strength at active sites. Additionally, the varying NV types also fine-tuned the adsorption energy, with the graphitic NV showing the strongest PMS adsorption, thus creating a high energy barrier of 3.61 eV for subsequent reaction. Conversely, the catalysts with pyridinic and pyrrolic NVs presented moderate PMS adsorption energies, more conducive to ensuing reactions (Fig. 1e).The combined influence of F and N also manifested in interfacial charge transfer dynamics. As shown in Fig. 1f, post-fluorination, a greater amount of electrons transferred from PMS to NV (0.079 e− for NV-NC-2 vs 0.109 e− for NV-NFC-2). Charge density differential analysis revealed that both oxygen atoms in the PMS O–O bond tended to lose electrons, shortening the bond length from 1.509 Å in free PMS to 1.476 Å and 1.469 Å in NV-NC-2 and NV-NFC-2, respectively (Supplementary Table 1). Meanwhile, the O–H bond length became longer (0.9970 Å in NV-NC-2 and 0.9977 Å in NV-NFC-2) than that in the free PMS (0.9767 Å) (Supplementary Table 1), promoting the selective breaking of the O–H bond and forming SO5•− (Eq. 1). By comparing the three types of NVs, the catalysts with pyrrolic N exhibited the highest charge transfer (Fig. 1f and Supplementary Fig. 3), making them energetically the most favorable in the above reaction to form SO5•−. Once SO5•− was formed, it readily generated SO42− and 1O2 (Supplementary Fig. 4) via the rapid self-reaction of SO5•− (Eq. 2)30.$${NV}s+\,{{HSO}}_{5}^{-}\,\to \,{{{SO}}_{5}^{\cdot -}}^{*}+\,{H}^{+}+\,{e}^{-}({NVs})$$
(1)
$${2e}^{-}({NVs})+\,{{{SO}}_{5}^{\cdot -}}^{*}+{{SO}}_{5}^{\cdot -}\,\to \,{2{SO}}_{4}^{2-}+\,{{\,\!}^{1}O}_{2}$$
(2)
Furthermore, F and N’s regulation of electronic structure also impacted the adsorption of phenol on the catalyst. As shown in Supplementary Figs. 5 and 6, phenol tended to be adsorbed at the F-C site. The introduction of F induced a change in the electronic structure, which significantly enhanced the phenol adsorption energy of the catalyst from −0.40 ~ −0.21 eV to −3.81 ~ −1.72 eV (Fig. 1g). The different types of NVs also provided fine tuning of the phenol adsorption energy, aligning with the trends observed in PMS adsorption (Fig. 1g). This simultaneous modulation of PMS and phenol adsorption energies underscored the synergy between the NV catalytic site and F-C adsorption site, significantly enhancing mass transfer and the utilization efficiency of 1O2 (Supplementary Fig. 7).MMT-assisted synthesis of the dual-site catalystWith the help of the rational design aided by theoretical simulation, we successfully synthesized a dual-site catalyst engineered for the efficient production and neighboring utilization of 1O2. In this catalyst, N and F work in concert to shape the electronic framework of NVs, which, alongside F-C Lewis-acid sites, form the core of our synergistic adsorption-reaction mechanism. However, prior challenges in F-doping limited the efficient synthesis of such dual-site catalysts31,32,33. Our method leverages MMT-assisted pyrolysis (NFC/M), which we experimentally validated to ensure controllable synthesis (Fig. 2a). Also, control samples (NFC, NC/M, FC/M, and NFC/SiO2) were prepared with some changes (details are given in “Methods”). The NFC/M catalyst showcased a unique 3D structure composed of nanosheets, greatly enhancing surface area and pore volume compared to NFC and NC/M (Fig. 2b, c and Supplementary Figs. 8–10 and Supplementary Table 2). During synthesis, MMT acted as an in-situ hard template and was subsequently etched away by F-rich gas released from the decomposition of polytetrafluoroethylene (PTFE) (Supplementary Fig. 11 and Supplementary Table 3)34, optimizing the catalyst’s morphology and increasing the accessibility of its active sites. The resultant NFC/M consisted of randomly orientated graphitic carbon with uniform elemental distribution of C, N, and F (Fig. 2d, e and Supplementary Fig. 12b), mirroring the FC/M benchmark (Supplementary Fig. 13). In contrast, NFC synthesized without MMT exhibited a highly localized distribution of F (Supplementary Fig. 12d). These observations underscore MMT’s crucial role in the controlled synthesis of the dual-site catalyst, balancing the decomposition kinetics of C, N, and F precursors during pyrolysis and catalyzing N and F co-doping.Fig. 2: Morphology and structure of NFC/M.a Illustration of preparation process. b SEM image of NFC/M. c TEM image of NFC/M. d HRTEM image of NFC/M. e HAADF-STEM image and corresponding EDS elemental mapping of NFC/M.The removal of MMT coincided with the emergence of NVs, characterized by diminished triazine units and changes in vibrational modes (Fig. 3a)22. The graphitic carbon structure of NFC/M stood distinct from the C3N4 structure of NC/M35 (Supplementary Figs. 14 and 15). The characteristic D and G bands of graphite carbon appeared at 1351 and 1558 cm−1, respectively (Fig. 3b). The ID/IG ratio of NFC/M, standing at 1.20, surpassed those of FC/M (0.74) and NFC (1.07), indicating a higher degree of defects due to the NV formation35. This is further evidenced by a central shift in the D and G bands in NFC/M, suggesting localized electron distributions36. In addition, the increased bulk phase C/N atomic ratio of 1.47 (vs. 0.60 for NC/M and 0.91 for NFC) reaffirmed the presence of NVs in NFC/M (Fig. 3c), aligning closely with the surface measurements (Supplementary Fig. 16a, b). Predominantly pyridinic and pyrrolic NV speciation was identified (Fig. 3d and Supplementary Fig. 17)37. Notably, NFC displayed a robust electron paramagnetic resonance (EPR) signal of unpaired electrons, with fewer NVs than NFC/M (Fig. 3b, c and Supplementary Fig. 16), indicating disruptions of its graphite carbon structure and increased carbon radical formation. These findings highlight the protective role of MMT during the synthesis of NFC/M. The decomposition of melamine created a reductive atmosphere, facilitating the formation of corrosive byproducts like HF (Supplementary Fig. 18), which was sequentially consumed or buffered by the SiO2 and Al2O3 layers within the MMT template (Supplementary Figs. 19 and 20), effectively mitigating the corrosive byproducts from PTFE decomposition. Such a protective mechanism shields the carbon matrix from chemical etching, thus promoting the selective formation of pyridinic and pyrrolic NVs34. This interplay underscores the importance of MMT in safeguarding the structure of the catalyst during synthesis, ensuring the stability and functionality of the resulting NFC/M.Fig. 3: Electronic and atomic structure of NFC/M.a FTIR spectra. b Raman spectra. c C/N molar ratio based on elemental analysis results. Data are presented as mean values ± SD (n = 3). d EPR spectra measured at room temperature. e High-resolution XPS spectra for F 1 s in NFC/M. f EIS spectra. g–i Normalized XAS spectra at the C K-edge, F K-edge, and N K-edge of different catalysts. a.u., arbitrary units. Source data are provided as a Source Data file.Abundant F-C Lewis acid sites were confirmed in NFC/M with a molar ratio of F reaching up to 4.17% (Fig. 3e and Supplementary Table 3)35. The strong electronegativity of F greatly enhanced the charge transfer capabilities of NFC/M, surpassing those of NFC, NC/M, and FC/M (Fig. 3f and Supplementary Figs. 21 and 22). Soft X-ray absorption near-edge structure (XANES) analysis revealed the intricate electronic structure of NFC/M, with three distinct peaks corresponding to π* C–C, π* C–N, and σ* C–C/C–F orbitals evident in the C K-edge spectra of NFC/M, distinguishing it from the C3N4 signal observed in NC/M (Fig. 3g)38. In contrast to NFC, the intensity of the σ* bands of NFC/M considerably increased relative to that of the π* bands for C, and the π* C–C peak shifted toward a higher energy, indicative of increased electron localization fostered by the proliferation of C–F bonds. This shift resulted in a decreased electron density around carbon atoms39, as confirmed by the F K-edge spectra (Fig. 3h). A significant peak at 694.8 eV denoted the C–F bond in NFC/M, its upward shift suggesting an elevated oxidation state of fluorine atoms, thereby enhancing electron density. Additionally, a prominent peak at 699.4 eV indicated F–Cx–N interactions among uniformly distributed F and neighboring N atoms in NFC/M, contrasting with NFC. The N K-edge showcased a notable decline in the intensity of the π* band relative to the σ* band for N in NFC/M (Fig. 3i), highlighting the prevalence of pyridinic and pyrrolic nitrogen vacancies40,41, echoing the X-ray photoelectron spectra (XPS) analysis (Supplementary Fig. 17)39. These findings not only confirm the creation of pyridinic/pyrrolic NVs and F–C Lewis-acid sites through MMT-mediated pyrolysis but also demonstrate how the synergistic interplay between N’s near-range effects and F’s long-range influence optimally modifies their electronic structures, crucial for the dual-site adsorption-reaction model in NFC/M.Exclusive 1O2 production from PMS activation on NFC/MInspired by the well-defined geometric and electronic structure of NFC/M, the exclusive 1O2 production was subsequently assessed by using multiple approaches. EPR spectroscopy reveals that using 2,2,6,6-tetramethylpiperidine (TEMP) as the 1O2 trapper resulted in a notable triplet peak signal of 2,2,6,6-tetramethylpiperidine-N-oxyl radical (TEMPO). This signal greatly diminished with the introduction of NaN3, a known 1O2 quencher without impacting PMS concentration, indicating a 1O2-specific reaction pathway (Fig. 4a and Supplementary Figs. 23a and 24)1,5. Notably, TEMPO signal intensity within the NFC/M system was substantially higher than that in the control groups and continued to rise during the reaction, showcasing its superior 1O2 production capabilities6. When employing 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) to trap potential radicals, no characteristic EPR signals indicative of •OH, SO4•−, and O2•− were observed, but only the 1O2-mediated 5,5-dimethyl-1-pyrrolidone-N-oxyl (DMPOX) signal was detected, as confirmed by the NaN3 probe (Supplementary Fig. 23b–d)5, further corroborating the selective generation of 1O2. High-resolution mass spectrometry supports these findings by identifying the formation of DMA-O2 from the interaction of 9,10-dimethylanthracene (DMA) with 1O23,42, confirming substantial 1O2 activity (Fig. 4b, c and Supplementary Figs. 25–27).Fig. 4: Performance and mechanism for selective 1O2 production from PMS activation on NFC/M.a EPR spectra in different systems using TEMP as trapping agents. b Excitation and emission spectra of DMA in 50.0 wt.% acetonitrile in H2O from treatment by NFC/M system. The excitation was recorded from 300 to 420 nm with λem = 428 nm. c HR-MS chromatogram of typical DMA-O2 from DMA oxidation in NFC/M system. d Quenching effects of the different scavengers on phenol degradation. e Qualitative and quantitative analyses of reactive species in NFC/M-PMS system. f Degradation kinetic of phenol under different atmospheres. g EPR spectra for the 1O2 detection in the presence of TEMP and BQ. h In situ Raman spectra. i In situ EPR spectra of NFC/M reaction systems. a.u., arbitrary units. Data for (d) and (f) are presented as mean values ± SD (n = 3). Source data are provided as a Source Data file.Subsequent scavenging experimental results highlight the pivotal role of 1O2 in pollutant degradation (Fig. 4d). The introduction of radical-quenching agents like methanol (MeOH), ethanol (EtOH), and tert-butanol (TBA) barely influenced phenol and bisphenol A (BPA) degradation, suggesting the negligible involvement of the other reactive species (Supplementary Fig. 28a–f)1,5. Conversely, specific 1O2-quenching agents, TEMP, NaN3, and furfuryl alcohol (FFA), at any concentration drastically inhibited pollutant degradation with the effect intensifying concomitant with the scavengers’ concentration, highlighting the dominant role of 1O2 in the NFC/M-mediated Fenton-like catalysis (Supplementary Fig. 28g-i). Additionally, replacing water with deuterium oxide (D2O) to extend the 1O2 lifespan moderately enhanced pollutant degradation, further conforming its key role in these reactions (Supplementary Fig. 29)1. No pollutant removal was observed in the galvanic oxidation system (GOS) with either blank or NFC/M-modified carbon electrodes, indicating negligible catalyst-mediated electron transfer from pollutant to oxidant (Supplementary Fig. 30)43. Therefore, 1O2 was the sole active species within the NFC/M-PMS system (Fig. 4e), and responsible for the pollutant degradation.The mechanistic insights into 1O2 production were further investigated. Despite possible involvement of dissolved oxyge14, the introduction of nitrogen or oxygen gas did not affect phenol degradation, dismissing these as sources for 1O2 formation (Fig. 4f and Supplementary Fig. 31). This result confirms that PMS was the sole source of 1O2. The absence of O2•− detection and an increased 1O2 signal in the presence of benzoquinone (BQ, a scavenger for O2•−) further confirm that 1O2 was not derived from O2•− intermediates (Fig. 4g and Supplementary Figs. 32, 33)14,30.In-situ Raman spectroscopy dissects the interactions between PMS molecules and the catalyst’s active site. The peaks at 1060 and 884 cm−1, corresponding to the vibrational modes of SO3− and the O–O in PMS (H–O–O–SO3−), and a peak at 980 cm−1 associated with the symmetric stretch of S=O bonds in SO42−, demonstrated a swift conversion of PMS to SO42− following the activation by NFC/M (Fig. 4h)44. This transformation, along with a noticeable shift of the O–O peak to 877 cm−1 post-interaction, suggests the decreased electron density and electron transfer from PMS to NFC/M23. This was supported by the consistent pollutant-independent PMS decomposition results (Supplementary Fig. 24).Electrochemical tests further validated that PMS served as the electron donor for 1O2 production. Chronoamperometric measurements showed a significant current increase upon PMS addition, and the open circuit potential (OCP) of the glassy carbon electrode coated with NFC/M rose immediately, indicating the strong surface interaction via electron transfer (Supplementary Fig. 34). However, phenol injection did not alter the current, signifying no electron interaction between phenol and PMS or the catalyst. A slight decrease in OCP could be attributed to the phenol adsorption on the NFC/M surface, possibly blocking active sites or modifying surface properties. Linear sweep voltammetry (LSV) analysis further confirms the electron transfer from PMS to the active sites, as evidenced by the increased current density at the NFC/M electrode (Supplementary Fig. 35)30.In in-situ EPR analysis, NVs were identified as key active sites. Compared to the stable unpaired electron signal on NFC without NVs (Supplementary Fig. 36), the presence of PMS led to a significant reduction in EPR signal intensity from NVs in NFC/M (Fig. 4i), indicating effective electron trapping. The signal initially decreased sharply and then stabilized, suggesting that NVs initially acted as electron acceptors to activate PMS, and subsequently as electron donors for 1O2 production, quickly reaching a dynamic equilibrium (Fig. 4i and Supplementary Fig. 37). These findings, aligning with our DFT simulations, provide robust experimental support for the mechanism of selective 1O2 production through NFC/M-activated PMS (Eqs. (1) and (2)).Neighboring 1O2 utilization on NFC/M for water decontaminationDriven by the exclusive selectivity and large production of 1O2, NFC/M exhibited superior Fenton-like performance for pollutant degradation. BPA was completely degraded within 2.0 min by NFC/M-PMS system, while no BPA removal was observed within 20.0 min by PMS or NFC/M alone (Fig. 5a). Compared to control samples, NFC/M showcased remarkable reactivity towards phenol degradation, characterized by high reaction kinetics and enhanced mineralization efficiency (Fig. 5b and Supplementary Figs. 38–40). The reactivity of the Fenton-like process correlated strongly with the concentration of 1O2 (Figs. 4a and 5b). In-situ Raman spectra further disclosed NFC/M’s significant phenol enrichment capability, bolstering 1O2 utilization (Fig. 4h). Phenol degradation followed pseudo-first-order kinetics, with NFC/M exhibiting the highest observed reaction rate constant (kobs) of 0.58 min−1, significantly surpassing that of NFC, NC/M, and FC/M by 58.1, 290.5, and 290.5 times, respectively (Fig. 5c). Such notable disparity in reactivity also underscores that residual Si, Al, and Na within the catalyst did not contribute to its catalytic activity (Supplementary Fig. 41). Deeper insights into the catalytic activity were obtained by assessing the impacts of NVs and F content on NFC/M (Supplementary Figs. 42–45). The catalyst’s performance was closely linked to these concentrations, which crucially influenced its surface electronic structure. Specifically, the catalytic activity correlated with the variations in NV concentration, initially increasing and later diminishing (Supplementary Figs. 43 and 44). This observation supports our catalyst design considerations that the electronic structure engineering of NV drastically boosted PMS activation, resulting in excellent performance facilitated by the synergistic effects of dual-site interactions.Fig. 5: Fenton-like performance via synergistic dual-site model.a BPA adsorption and oxidation via PMS activation on NFC/M. b, c Reactivity comparison between NFC/M and references for phenol degradation. d, e Degradation profiles (d) and removal efficiency after 20 min reaction (e) of different pollutants in NFC/M system. f Comparison of degradation kinetics with recently reported PMS activation processes. The green, blue, and gray symbols represent single-atom catalysts, metal-free catalysts, and other metal-based catalysts, respectively. Among them, the symbols with purple dashed circles indicate catalysts that predominantly generate 1O2 as the ROS. See Supplementary Table 4 for details. g Reusability of NFC/M. h Influence of pH on phenol degradation in NFC/M system. i Phenol degradation kinetics with interference of ions and humic acids. j The schematic illustration of the continuously flow filter. k Continuous operation test of BPA degradation in the continuously flow filter. Inset (i), photograph of NFC/M @ carbon felt. Inset (ii), SEM images to illustrate the NFC/M catalyst coated on the carbon felt. Reaction conditions for (a–i): [catalyst] = 0.2 g·L−1, [PMS] = 0.65 mM, [pollutant] = 20.0 mg·L−1 (4-CP = 25.7 mg·L−1; 2, 4, 6-TCP = 39.5 mg·L−1; RhB, MB = 50.0 mg·L−1), initial pH 7.0 (if not adjusted), Temp. = 20.0 ± 2.0 °C; for (k), flow rate = 3 mL·min−1, catalyst loading = 0.1 g (if needed), [PMS] = 0.65 mM, [BPA] = 5.0 mg·L−1, HRT = 2.6 min, Temp. = 20.0 ± 2.0 °C. Data for (a–e) and (g–i) are presented as mean values ± SD (n = 3). Source data are provided as a Source Data file.Further tests on the degradation of an array of nine additional pollutants validated the broad-spectrum efficacy of NFC/M (Fig. 5d). Within 20 min, various phenolic contaminants, pharmaceutical, and personal care products, as well as dyes, including p-Chlorophenol (4-CP), 2,4,6-Trichlorophenol (2,4,6-TCP), sulfamerazine (SMZ), sulfanilamide (SA), ofloxacin (OFX), rhodamine B (RhB), and methylene blue (MB), could be removed by nearly 100% (Fig. 5e). Sulfamethoxazole (SMX) was reduced by approximately 70% (Fig. 5e), while nitrobenzene (NB), a compound with electron-withdrawing properties, showed minimal degradation (Supplementary Fig. 46). These findings highlight NFC/M’s selective efficacy against electron-rich pollutants, facilitated by the mild redox potential of 1O21,45. The performance of NFC/M surpassed that of many advanced catalysts, demonstrating the highest capacity for selective 1O2 production and remarkable catalytic stability in pollutant degradation (Fig. 5f, g and Supplementary Table 4). Moreover, NFC/M exhibited considerable pH resilience, operating effectively across a range from 3.0 to 9.5, albeit with a slight performance dip at an initial pH of 10.8 due to surface repulsion and OH− quenching effects on 1O2 (Fig. 5h)46. Additionally, phenol removal was highly resistant to interference from humic acids (HA) and various coexisting ions at all tested concentrations (Fig. 5i and Supplementary Fig. 47), further underscoring its robustness in complex water matrices.The practical application of NFC/M in real-world scenarios was also examined, particularly in the treatment of industrial wastewater, where it achieved significant chemical oxygen demand (COD) removal efficiencies (Supplementary Tables 5 and 6). Over 50% COD removal was accomplished within 60 min in aniline production wastewater, while 48.2% of COD was removed from biochemical wastewater in the same timeframe. Considering the challenging nature of these industrial effluents, NFC/M’s performance was notably impressive.Exploring its application potential at a device level, NFC/M was integrated into a commercial carbon felt (45 mm diameter, 5 mm thickness) to construct a water-purification filter (Supplementary Fig. 48). The high surface area and porosity of carbon felt enhanced the treatment capacity of the filter46,47,48. Its effectiveness in continuous pollutant removal was assessed in a flow-through setup (Fig. 5j), where a solution containing 5 ppm BPA and 0.65 mM PMS was processed at a flow rate of 3 mL/min with a hydraulic retention time of just 2.6 min. Thanks to the abundant percolating channels and well-distributed NFC/M within the filter (Supplementary Fig. 49), it maintained over 86.8% BPA removal efficiency after 36 h of operation (815 bed volumes) (Fig. 5k). In contrast, BPA removal efficiency sharply decreased in filtration systems lacking NFC/M, primarily due to the saturated adsorption of the carbon felt. Overall, the NFC/M system showcased exceptional adaptability, durability, and catalytic performance in handling complex wastewater, affirming its viability for practical water purification scenarios.
Tuning electronic structure of metal-free dual-site catalyst enables exclusive singlet oxygen production and in-situ utilization
