Introduction

Since fluorine has the strongest electronegativity in the periodic table and is the second smallest atom after hydrogen, introducing fluorine and fluorine-containing groups into organic molecules can effectively improve their physicochemical properties, such as acidity, lipophilicity and stability1. The ubiquitous use of organofluorine chemicals poses a serious globally realistic risk to human health and ecosystems2,3. Representative compounds such as perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), listed under the Stockholm Convention on Persistent Organic Pollutants (POPs)4,5, demonstrate significant spatial and vertical distribution gradients in the Bohai and Yellow Seas6. These contaminants exhibit long-range atmospheric transport potential, aquatic mobility, pronounced bioaccumulation tendencies, and considerable ecotoxicological effects7.

Existing organofluorine treatment technologies are mainly divided into two categories: adsorption-separation and degradation-destruction. Activated carbon8,9,10, ion exchange resin11,12,13 and mineral materials14 are commonly used as adsorption-separation materials to achieve the enrichment and transfer of organofluorine, but they still face challenges such as difficulty in adsorbent regeneration, low removal rate, low selectivity, and risk of secondary pollution. On the other hand, conventional pyrolysis15,16,17, incineration18,19, ultrasonication20,21, plasma-based oxidation22, electrochemical degradation23,24, supercritical water oxidation25,26 and ultraviolet-initiated degradation27,28,29,30 have been developed. However, current destruction strategies have shortcomings, including high energy consumption, harsh conditions, the use of heavy-metal catalysts, and toxic side effects such as fluoride ions-induced emission and contamination31. Recently, Brittany Trang et al. reported a mineralization of PFOA through a sodium hydroxide–mediated defluorination pathway, which converted the ‘forever chemical’ PFOA to fluoride ions within 24 h in polar aprotic organic solvents32. Undoubtedly, this robust alkali condition is too harsh and dangerous for actual sewage treatment. Leveraging solar energy and the reactivity of Csp2-F and Csp3-F bonds33,34, solar-driven mineralization might offer milder alternatives to address the impending risk of the organofluorine contamination.

Solar-driven degradation is a promising method to actuate efficient cleavage of organofluorine pollutants owing to its merits of facile operation, high efficiency and low secondary pollution35. Some previous works reported PFOA was completely degraded into CO2 and F− on TiO2 photocatalyst under the irradiation of mercury lamp (mainly ultraviolet light)36, and the fluorinated liquid crystal monomer (LCM) pollutants were degraded under the ultraviolet/peroxy disulfate treatment37. However, these photodegradation systems were based on wide bandgap semiconductors (TiO2, In2O3, Ga2O3, BiOHP, etc) containing precious metal and heavy metal catalysts38, which can only use ultraviolet light (mainly 254 nm) and fail to utilize abundant sunlight39,40,41. Besides, the degradation of organofluorine releases harmful small molecule fragments or F ions, and does not fully realize the mineralization and harmlessness of pollutants. To date, few research has explored the mineralization of organofluorine pollutants containing different hybrid C-F bonds under visible light32. Therefore, there is an urgent need to develop visible light-responsive photocatalysts, which should demonstrate exceptional efficacy in circumventing the prohibitive thermodynamic constraints governing C-F bond cleavage in persistent organofluorine contaminants, as well as concomitant advantages in economic feasibility and ecological safety.

Mesoporous silica (mSiO2), which is similar to the main components of sand and environmentally friendly materials, has a large specific surface area to enhance the adsorption of organic pollutants and fluoride ion42,43. Meanwhile, cerium oxide (CeO2) is a strong oxidizing photocatalyst, which can produce highly active hydroxyl radical (•OH) species under visible light irradiation to destroy organic pollutants. Herein, we present a solar-powered mineralization strategy utilizing CeO2/mSiO2 heterojunction photocatalysts for complete defluorination of organofluorine contaminants. Subsequently, fluorinated e-waste octafluoro-4,4’-biphenyldiamine (OFB)44,45,46,47,48, fluoro-antibiotic fleroxacin (FLE)49,50,51, and perfluorinated surfactant PFOS27 were selected as representative organofluorine pollutants with typical Csp2-F and Csp3-F bonds to explore the visible-light removal performance (Fig. 1A). Notably, experimental results and density functional theory (DFT) calculations reveal that the destruction of Csp2-F and Csp3-F bonds by photogenerated •OH was thermodynamically superior to hydroxyl (OH−)-mediated defluorination under heating conditions (Fig. 1B). This work provides a feasible photodegradation technology for sustainable wastewater management in China, especially because emerging contaminants with ecological risks are not given enough attention.

Fig. 1: Schematic diagram illustrating the solar-driven mineralization of organofluorine pollutants.
Fig. 1: Schematic diagram illustrating the solar-driven mineralization of organofluorine pollutants.The alternative text for this image may have been generated using AI.
Full size image

A Fluorinated e-waste OFB, fluoro-antibiotics FLE and perfluorinated surfactant PFOS were mineralized by CeO2/mSiO2 photocatalyst. B Comparison of the destruction of PFOS via photogenerated •OH under solar and hydroxyl-mediated defluorination under heating. SEM (C) and HRTEM (D) images of CeO2/mSiO2, showing (311) lattice fringes of CeO2. E Photodegradation of PFOS under the illuminations of visible light (410-nm LED) and simulated sunlight (Xe lamp, AM1.5 G) in the presence and absence of CeO2/mSiO2 (2.0 g L−1). Photograph of the outdoor sunshine experiment (F) and the removal rate of PFOS by CeO2/mSiO2 catalyst after 5 days illuminations (G), along with the local monthly average temperature recorded over a one-year period. Data are denoted as mean ± s.d. (n = 3).

Results

Preparation and characterization of CeO2/mSiO2 nanocomposites

CeO2/mSiO2 nanocomposites with tunable CeO2 loadings (5–80 wt%) were fabricated via the sol-gel method and subsequent hydrothermal reaction (Supplementary Fig. S1). Under alkaline conditions, tetraethyl orthosilicate (TEOS) undergoes hydrolysis to form silicic acid (Si(OH)4), while CTAB micelles act as structural templates. Subsequently, silicate species condense on the surface of the micelles, forming a core-shell composite structure with the micelles as the core and silica as the shell. After removing the CTAB template via high-temperature calcination, mesoporous SiO2 (mSiO2) with a high specific surface area is obtained. Furthermore, introducing mSiO2 during the in-situ synthesis of CeO2 enables the uniform dispersion of cerium salts within the mSiO2 matrix. This approach effectively prevents the severe agglomeration of CeO2 nanoparticles that typically occurs during high-temperature calcination, thereby preserving their catalytic performance. As evidenced by scanning electron microscopy (SEM), the optimized 5%CeO2/mSiO2 composite demonstrates monodisperse spherical nanostructures with an average diameter of 50 nm (Fig. 1C). High-resolution transmission electron microscopy (HRTEM) analysis reveals distinct lattice fringes with a d-spacing of 0.31 nm, corresponding to the (311) crystallographic plane of CeO2, confirming successful surface decoration of CeO2 nanoparticles on the mesoporous silica matrix (Fig. 1D). Energy dispersive X-ray spectroscopy (EDX) elemental mapping (Supplementary Figs. S2, S3) verifies the homogeneous distribution of Ce, Si and O within the composite architecture. X-ray powder diffraction (XRD) patterns (Supplementary Fig. S4) exhibit progressive enhancement of characteristic CeO2 diffraction peaks of (111), (200) and (311) (JCPDS 34-0394) with increasing CeO2 loading, while pristine mSiO2 maintains an amorphous profile. The linear correlation between CeO2 content and peak intensity confirms controlled crystallinity modulation through synthesis parameter optimization.

N2 physisorption isotherms (Supplementary Fig. S5) demonstrate type-IV hysteresis loops, characteristic of hierarchical mesoporous structures. The 5%CeO2/mSiO2 composite retains exceptional surface area (884.7 m2 g−1) and pore volume (1.61 cm3 g−1), comparable to bare mSiO2 (913.7 m2 g−1, 1.65 cm3 g−1), with bimodal pore size distributions centered at 1.5 and 3.8 nm (micropores) and 30–60 nm (mesopores). Notably, these values surpass conventional SiO2 aerogels and nanospheres (Supplementary Table S1)42,52,53, attributable to the structural preservation during CeO2 integration. The superior performance stems from fundamental structural differences: CTAB-templated mSiO2 achieves precisely controlled 2–50 nm mesopores through micelle-directed sol-gel assembly, yielding exceptionally high surface area and uniform pore distribution. In contrast, conventional SiO2 aerogels exhibit macropores (tens-hundreds nm) with surface area primarily from nanoscale frameworks rather than optimized mesoporosity. The slight reduction in chiral mSiO2‘s surface area versus standard mSiO2 arises from chiral dopants perturbing CTAB micelle self-assembly during synthesis, reducing template uniformity. Crucially, our material’s mesopore density provides 3.2-folds more accessible active sites per unit volume than aerogels, explaining the enhanced performance metrics.

UV-Vis diffuse reflectance spectroscopy (Supplementary Fig. S6) reveals a redshifted absorption edge (λ > 430 nm) and reduced optical bandgap (2.86 eV) for 5%CeO2/mSiO2 compared to pristine mSiO2 (3.32 eV), indicating enhanced visible-light harvesting capability. Electrochemical impedance spectroscopy (EIS) Nyquist plots (Supplementary Fig. S7) demonstrate a relatively small semicircle compared with bare mSiO2, suggesting that the loading of CeO2 enhances the charge transfer efficiency of mSiO2. Density functional theory (DFT) calculations the total density of states (TDOS) of CeO2/mSiO2 shows increased electronic states near the Fermi level (Supplementary Fig. S8). This synergistic enhancement in charge separation and mobility originates from the type-II heterojunction formation at the CeO2/mSiO2 interface.

Photocatalytic degradation of organofluorine pollutants

The photocatalytic efficacy of CeO2/mSiO2 nanocomposites was systematically evaluated using three structurally distinct organofluorine contaminants: fluorinated e-waste octafluorobiphenyl (OFB, Csp²-F bonds with amino groups)44,45,46,47,48, perfluorooctanesulfonic acid (PFOS, Csp³-F bonds with sulfonic acid groups)27, and fluorinated antibiotic fleroxacin (FLE, hybrid Csp²-F/Csp³-F bonds containing carboxylic/amino functionalities)49,50,51 (Fig. 1A, Supplementary Fig. S9). Under visible light irradiation (410 nm LEDs, 50 mW cm−2), the optimized 5%CeO2/mSiO2 achieved 91.1 ± 3.2% OFB degradation within 6 h, significantly outperforming commercial TiO2 (52.0 ± 1.9%), pure CeO2 (69.6 ± 1.5%), and bare mSiO2 (11.2 ± 0.5%). Reaction kinetics analysis revealed pseudo-first-order rate constants (k) of 0.42 h−1 (OFB), 0.31 h−1 (FLE), and 0.18 h−1 (PFOS), correlating with bond dissociation energies of Csp²-F (544 kJ mol−1) versus Csp³-F (~452–489 kJ mol−1). Zeta potential measurements demonstrated pH-responsive surface charge modulation of CeO2/mSiO2, ranging from +0.68 mV (pH 3.3) to −1.74 mV (pH 11.0) (Supplementary Fig. S10). This electrostatic complementarity drives substrate-specific adsorption: OFB (pKa > 9) exhibited enhanced adsorption in alkaline condition via amino-CeO2 coordination, while PFOS (pKa < 1) achieved enhanced adsorption in acidic solution through sulfonate-silica interactions (Supplementary Fig. S11). FLE’s amphoteric nature enabled dual-mode adsorption across pH 5-9, maximizing degradation efficiency (97.7 ± 2.8%).

Comparative experiments were conducted under various illumination conditions: darkness, monochromatic visible light (3 W 410-nm LEDs, ~30 mW cm−2), and simulated sunlight (Xenon lamp with AM 1.5 G filter, 100 mW cm−2). In the dark, the catalyst has no degradation effect on PFOS (Fig. 1E), despite CeO2 being known for its strong oxidation ability to destroy hydrocarbons. However, in the presence of the CeO2/mSiO2 catalyst and light irradiation, whether from the 410-nm LED or simulated sunlight, the removal rate of PFOS exceeded 80% within 150 min (Fig. 1E), the removal rate of OFB exceeded 90% within 150 min, and FLE exceeded 90% within 50 min (Supplementary Fig. S12). For a horizontal comparison, PFOS, with its highly stable Csp3-F bonds, is the most difficult to degrade (well-known as ‘forever chemical’), while FLE with C-H bonds is the easiest to degrade. In the absence of catalyst, the degradation effect of organofluoride was significantly inhibited. Additionally, the impact of different anions and cations on organofluorine degradation was investigated (Supplementary Fig. S13). Among common anions, the CO32− anion inhibits the degradation of OFB and FLE, while cations such as Na+, K+ and Ca2+ have negligible effects on the degradation of organofluorides.

Solar-driven mineralization of organofluorine contaminants under outdoor illumination was carried out over a one-year period (Fig. 1F), with parallel experiments in May, August, November, and January, (9:00 a.m. to 4:00 p.m, 7 h per day). At the measurement site (Shantou, Guangdong, China; 23.3541°N, 116.6820°E), solar irradiance reached 58.5 mW cm−2 at 12:30 local time. Given that PFOS exhibits an extremely slow natural degradation rate (half-life of 4.7)54, no significant photolysis occurred within 5 days of direct outdoor sunlight exposure. In contrast, after introducing the CeO2/mSiO2 photocatalyst, 26.8 ± 6.3% of PFOS underwent destruction within 5 days under sunlight illuminations, and the degradation efficiency fluctuated slightly with season and temperature (Fig. 1G). To our knowledge, this is the first report on the efficient removal of PFOS in outdoor environments by sunlight, which is expected to be applied to the environmental treatment of organofluorine pollutants. Additionally, 83.5 ± 2.1% of OFB and 97.7 ± 2.8% of FLE underwent photodegradation under outdoor sunlight for one-day irradiation (7 h), which was a 2.58- and 1.51-fold enhancement of the photolysis without a catalyst (Supplementary Fig. S14). This solar-driven approach achieves degradation efficiency comparable to conventional advanced oxidation processes (AOPs) such as electrochemical oxidation, Fenton-based methods, plasma treatment, and sonochemical degradation (Supplementary Table S2), while operating under milder conditions suitable for long-term remediation of perfluorinated compounds in outdoor environments.

The intermediates and final products of organofluorine photodegradations were identified using liquid chromatography-mass spectrometry (LC-MS) and gas chromatography (GC) analysis (Fig. 2, Supplementary Figs. S15–S19). As shown in Fig. 2A, the signal changes of each ion fragments of OFB were monitored during 200 min of continuous illumination to elucidate its photodegradation pathways. The signal peak of OFB at m/z 329 vanished completely after 40 min of illumination (Supplementary Fig. S15A). Concurrently, the signal of a2 at m/z 318 disappeared at a retention time of 6–8 min after 40 min of illumination, but reappeared at a retention time of ~2 min, then gradually weakened and ultimately disappeared within 200 min (Supplementary Fig. S15B). Similar trends were observed in the figures of m/z 275, 218, 177, 120 and 90, which exhibited new peaks at retention times of 0.77, 5.8–6.2, 6–7, 0.68 and 0.71 min, respectively (Supplementary Fig. S15C–G). Integrating molecular weight and retention time, the ion peak signals of m/z 120, 90, 85 and 64 were identified as C3 or C2 intermediates resulting from OFB photodegradation (Supplementary Fig. S15F–I). More intriguingly, after 160 min of illumination, a fragment signal peak of m/z 60 suddenly increased at the retention time of 0.74 min, which was confirmed to be urea by comparison with a standard sample (Supplementary Fig. S15J). Additionally, the indophenol blue method was employed to track the degradation of OFB, ultimately yielding 0.1623 μg mL−1 of ammonia (Supplementary Fig. S16), while GC analysis confirmed that OFB photodegradation ultimately produced CO2 (Supplementary Fig. S17A). Therefore, the intermediates of ammonia and CO2 could be transformed into urea under continuous illumination.

Fig. 2: Detection of intermediate products and proposed degradation pathways.
Fig. 2: Detection of intermediate products and proposed degradation pathways.The alternative text for this image may have been generated using AI.
Full size image

The ion fragments intensities of OFB (A), FLE (B) and PFOS (C) during photodegradation by the CeO2/mSiO2 catalyst, and the rational degradation paths incorporating intermediates detected by LC-MS spectrometry and GC analysis for OFB (D), FLE (E) and PFOS (F), respectively. Data are denoted as mean ± s.d. (n = 3).

The decomposition of FLE is anticipated to initiate from the functional groups at either the N-substituted piperazine or N-substituted pyridine, which can be divided into two pathways and yield intermediates of b1–b10 with m/z of 343, 286, 240, 258, 323, 355 and 271, respectively (Fig. 2E). As illustrated in Fig. 2B, the time-varying signals confirm that FLE decomposed to b1 and b5’ within 40 min of illumination (Supplementary Fig. S18). Subsequently, they transformed into benzene or pyridine derivatives with m/z of 188, 184 and 139, followed by smaller organic molecule fragments of oxalic acid and N-methyl amino acetic acid. Notably, the detected pyridine derivative intermediate of b7 with m/z of 139 has not been previously mentioned in the reported degradation mechanism of FLE49,51,55. Due to the low electron cloud density of the pyridine structure, b7 intermediate is more resistant to oxidants than other benzene derivatives and is challenging to degrade completely. Ultimately, FLE was decomposed into NH3 (0.5122 μg mL−1) (Supplementary Fig. S16C) and CO2 (Supplementary Fig. S17B).

The degradation of PFOS initiates with the elimination of the sulfonic acid group to form PFOA, followed by the decarboxylation-hydroxylation-elimination hydrolysis pathway to shorten the carbon chain (Fig. 2F). Under continuous illumination, intermediate species with m/z of 413, 263, and 113 were produced, which are attributed to PFOA, perfluoropentanoic acid (PFPeA), trifluoroacetic acid (TFA), respectively (Supplementary Fig. S19). Within 360 min of illumination, the concentration decay of PFOS and the intermediates of PFOA and TFA were statistically analyzed, confirming that PFOS was continuously converted into PFOA and eventually reduced the carbon chain to TFA (Fig. 2C). Additionally, the final product of CO2 was detected by GC analysis (Supplementary Fig. S17C). The total organic carbon (TOC) variations in the CeO2/mSiO2 catalyst systems with OFB (11.4 μM, room temperature saturated solution), FLE (50 μM), and PFOS (80 μM) were measured before and after light irradiation. Initially, their TOC values were 2.40, 9.43, and 7.11 mg L−1, close to theoretical values. After prolonged irradiation, the TOC values decreased to 1.5–1.8 mg L−1, with FLE and PFOS showing TOC reductions of 80.2% and 74.0%, reflecting the complete degradation of these organic fluorine pollutants (Supplementary Fig. S20).

DFT calculations of Csp2-F and Csp3-F cleavage

Radical quenching experiments and electron paramagnetic resonance (EPR) analysis were conducted to elucidate the active species involved in the photodegradation of organofluorine. Upon illumination, the photogenerated electron (e−) and hole (h+) are transferred to the surface of CeO2/mSiO2 semiconductor, where e- can be captured by O2 to form superoxide ion (•O2−) and h+ can be trapped by water to produce hydroxyl radical (•OH). In the EPR measurements, characteristic signals of •OH and •O2− were detected on the CeO2/mSiO2 catalyst under visible light irradiation, with intensities that increased alongside the extension of irradiation time (Supplementary Fig. S21). After purging argon to remove O2, the CeO2/mSiO2 photocatalyst could still degrade OFB, FLE and PFOS under irradiation, thus excluding •O2- as an indispensable active species for organofluorine degradation (Supplementary Fig. S22). The photodegradation of organofluorine pollutants was significantly reduced by adding reagents to quench •OH and h+, indicating that •OH plays a crucial role in the photodegradation process (Supplementary Fig. S23). DFT calculation results reveal the spatial structure and electrostatic potential (ESP) distributions of OFB, FLE and PFOS (Supplementary Fig. S24), in which the carbon atoms of the C-F bonds are in a charge-deficient state and are therefore more susceptible to attack by hydroxyl radicals or hydroxide ions.

To further elucidate how active species participate in the cleavage of Csp2-F bonds, fluorobenzene (Ph-F) was employed as the primary computational model. The feasibility of defluorination through nucleophilic elimination (Path I), oxidation pathways (Path II), and reduction pathways (Path III) was assessed using DFT theoretical calculations (Fig. 3A, Supplementary Fig. S25). Paths I and II are thermodynamically extremely difficult to occur Csp2-F bond breaking due to the high-energy barrier of benzyne intermediate (85.44 kcal mol−1) and cationic radical (+216.47 kcal mol−1). In contrast, the energy barrier for PhF to gain electrons to form anionic radicals in Path III is only +12.58 kcal mol−1, and the subsequent reaction with •OH is also a thermodynamically favored process. After the introduction of the strong electronegative fluorine on the conjugated aromatic ring, the π electron cloud is reduced, which facilitates electron acceptance. Consequently, the elimination of Csp2-F on the aromatic ring can be achieved by capturing photogenerated electrons, then reacting with •OH and dissociating fluoride ions. Afterward, the defluorination of OFB follows a similar Path III, losing four F atoms at the ortho-positions of the amino groups to form a1 intermediate, which is thermodynamically preferred, and the product was confirmed in LC-MS (Fig. 3B). In contrast, the fluorine atoms at the meta-positions are inert due to their significant steric hindrance. Subsequently, the a1 intermediate could undergo a C-C bond break (bond energy 130.95 kcal mol−1) to produce one-ring a3 intermediate, or lose electrons to form a2 intermediate with quinone structures and followed by C-C bond cleavage to form one-ring a4 intermediate (Supplementary Fig. S26). By comparing the C-C bond energies, the destruction of the benzene ring structure via the quinone intermediate is the most thermodynamically favorable path (Supplementary Table S3). Consequently, organofluorine pollutions in e-waste containing Csp2-F bonds could be activated by receiving photogenerated electrons and reacting with •OH to form phenolic hydroxyl structures, which can then be oxidized to carbonyl-containing quinone structures and gradually dissociate into small molecular fragments.

Fig. 3: Proposed defluorination mechanisms for Csp2-F and Csp3-F bonds by OH− or •OH.
Fig. 3: Proposed defluorination mechanisms for Csp2-F and Csp3-F bonds by OH− or •OH.The alternative text for this image may have been generated using AI.
Full size image

A Energy profiles of three paths for Ph-F conversion to Ph-OH. B DFT calculated free energy change when OFB gradually eliminates Csp2-F bonds, with specific C-C bond energies in brackets for the intermediates of a1, a2 and a2’. Degradation of PFeSA (C) and PFOS (D) by OH− under heating and •OH under solar irradiation, with the predicated intermediates and their free energy changes (kcal mol−1).

The cleavage of Csp3-F was analyzed using perfluoroethane sulfonic acid (PFeSA) as the primary computational model. According to the recent work in Science32, heating PFOA and NaOH in polar aprotic organic solvents could effectively remove Csp3-F via hydroxyl-mediated defluorination. Here, the destruction of Csp3-F in perfluorinated compounds by heated OH- and photogenerated •OH was compared (Fig. 3C, Supplementary Fig. S27). Under heating conditions, the direct removal of sulfur trioxide (SO3) or the nucleophilic substitution of SO3− by OH− are thermodynamic energy-consuming processes. A plausible pathway is to produce an alkyl carbon anion CF3CF2− and a hydrogen sulfate anion (HSO4−), with a free energy change of −18.52 kcal mol−1, which represents a thermodynamically favorable reaction pathway. However, this carbon anion encounters an energy barrier exceeding 100 kcal mol−1 when it directly releases an F− to form an olefin structure. Then, it tends to form an extremely stable CF3CF2-H structure by interacting with the solvent water molecule, thereby significantly hindering the subsequent defluorination process. Conversely, the photogenerated •OH could react with PFeSA to produce a carbon-centered radical CF3CF2• or an alcohol of CF3CF2OH with free energies of −17.64 and −71.42 kcal mol−1, respectively, which are thermodynamically feasible reactions. Furthermore, the α-difluorool (-CF2OH) structure rapidly and spontaneously hydrolyzes to acyl fluoride (-COF) and subsequently to carboxylic acid (-COOH), releasing two HF molecules.

As depicted in Fig. 3D, the degradation of PFOS by OH– and •OH under heating and illumination conditions was compared. In the heating alkaline solution, the sulfonate anion (INT1) formed by PFOS dissociation is attacked by OH−, resulting in the production of HSO4− and a carbon anion intermediate (INT2). However, INT2 tends to further evolve into superhydrophobic and chemically stable intermediates INT3, INT4 and INT5, making subsequent defluorinations difficult to proceed (Supplementary Fig. S28). In contrast, PFOS can be directly converted into sulfonate radicals (INT6) by photo-generated •OH, and then lose the SO3 fragment to form a carbon radical •CF2R (ITN7), and subsequently react with •OH to form an intermediate (INT8) with an active RCF2OH group. The free energy changes for these continuous reaction steps are −5.48, −12.11 and −98.40 kcal mol−1, respectively, indicating thermodynamically favorable processes. Subsequently, the α-difluorool (-CF2OH) group undergoes hydrolysis reactions to form geminal diol (INT9), acyl fluoride (INT10) and finally carboxylic acid groups (INT11). Following this, the generated perfluorooctanoic acid (PFOA) can be gradually reduced by a difluoromethylene unit (CF2) using a similar reaction path, with free energy changes of −186~−188 kcal mol−1 (Supplementary Fig. S29). The free energy changes of the stepwise chain-shortening reactions exceed the free energy change of the initial degradation of PFOS to PFOA, confirming the super stability and difficulty in initiating the destruction of PFOS. In conclusive, under mild illumination conditions, photogenerated •OH offers a favorable mechanistic pathway for the destruction of Csp3-F bonds.

The fluoro-antibiotics of FLE is speculated to have three parallel degradation pathways: deflourination of Csp2-F bonds, piperazine ring cleavage, and aromatization involving N-fluoroethyl elimination. The plausibility of the intricate degradation path of FLE was confirmed by energy barrier and bond energy calculations, with the lowest starting energy barrier (2.15 kcal mol−1) for the deflourination of Csp2-F bonds (Supplementary Fig. S30). Similar to the previous Csp2-F bond destruction process, the aromatic ring of FLE accepts an electron and then reacts with •OH to produce phenolic hydroxyl structure, achieving defluorination of Csp2-F bonds, which is thermodynamically permitted. In the second pathway, piperazine undergoes oxidative ring opening, in which the nitrogen atom in piperazine loses electrons to form an imine cation, and then hydrolyzes, leading to the breaking of the C-N bond. This pathway has been speculated and mentioned in the literature47,49,51,55, even though the reaction energy barrier for the initial electron loss process is high. In the third pathway, FLE eliminates the β-fluoroethyl group and then rearranges protons to form a benzopyridine skeleton structure with a low free energy change of 2.15 kcal mol−1. These paths are intertwined and occur simultaneously or successively, allowing FLE to be degraded into various small fragments, as confirmed by LC-MS analysis. Considering the C-N bond energies in N-ethyl and piperazine, the C-N bond energy in ethyl is lower (100 kcal mol−1) than that of C-N bonds in piperazine (151.58 and 141.96 kcal mol−1) (Supplementary Table S4), making the aromatization pathway feasible, which is consistent with the results of LC-MS analysis. After aromatization, the C-N bond energy of piperazine decreases to 138.14 and 138.03 kcal mol−1, which is beneficial to the subsequent C-N dissociation process. In addition, the destruction of 2-fluoroethanol fragment by •OH is thermodynamically favorable.

Defluorination of organofluorine

Unexpectedly, the concentration of F− in the solution did not increase significantly after the complete photodegradation of OFB, FLE and PFOS using CeO2/mSiO2 photocatalyst (Fig. 4), which is advantageous compared to the reported systems that release F− and CO241,56. In the comparative experiments, the concentration of F− increased to 1.05 μg mL−1 after 120 min of illumination using CeO2 as photocatalyst, which was close to the theoretical value (1.18 μg mL−1) of organofluorine pollutants (Fig. 4A). This result indicates that the generated F− is adsorbed by the CeO2/mSiO2 catalyst, as mesoporous SiO2 is a potential adsorbent for removing free F−42,43. EDS mapping shows that the F element was primarily distributed on the edges of the catalyst particles (Supplementary Figs. S31–S33). Notably, the contents of F element in the CeO2/mSiO2 catalysts increased to 1.55%, 2.48% and 2.20% after the photodegradation of PFOS, OFB, FLE, respectively. In FTIR spectra, the characteristic signals of organofluorine, attributed to the stretching vibration (ν) of C-F, N-H, aromatic ring, C-N, and the bending vibration (δ) of N-H, are obviously attenuated or even disappeared after illumination. Interestingly, CeO2/mSiO2 with organofluorine after illumination shows a weak signal at a lower wavenumber than C-F, which is assigned to the formation of Si-F bonds57. The XRD of CeO2/mSiO2 soaked with sodium fluoride (NaF) solution exhibits new diffraction peaks at 38.83° and 77.73°, indicating the chemical adsorption of F− on the catalyst. Besides, the rationality of the chemical adsorption of mSiO2 and F ions was confirmed by DFT calculations, which shows a free energy change of −24.20 kJ mol−1 from the conversion of Si-OH to Si-F (Fig. 4C). From the perspective of reaction equilibrium, the conversion of soluble F- into insoluble Si-F states is conducive to the degradation of organofluorine. Consequently, CeO2/mSiO2 not only achieves the destruction of organofluorine under sunlight irradiation but also does not release harmful F−, realizing the mineralization of organofluorine pollutants.

Fig. 4: Exploration of the adsorption of F- by CeO2/mSiO2 catalyst after photodegradation.
Fig. 4: Exploration of the adsorption of F- by CeO2/mSiO2 catalyst after photodegradation.The alternative text for this image may have been generated using AI.
Full size image

A The concentration change of F− during the photodegradation of OFB. B Theoretical and detected F− concentrations after completely photodegradation of OFB (11.4 μM), FLE (50 μM) and PFOS (10 μM) with irradiation times of 2, 2 and 6 h, respectively. C Free energy change of the conversion of Si-OH into Si-F. XPS survey (D) and high-resolution Si 2p (E) and F 1s (F) of CeO2/mSiO2 before and after degradation of PFOS, taking CeO2/mSiO2 soaked in NaF solution as a reference.

XPS measurements were conducted to compare the elemental compositions and valence states of CeO2/mSiO2 catalyst before and after the photodegradation of PFOS, OFB and FLE (Fig. 4D–F, Supplementary Figs. S33–S35). The XPS survey spectra show the characteristic peaks of Si, O and Ce elements, while the F characteristic peak (~685 eV) emerges after photodegradation (Fig. 4D). In the high-resolution XPS F 1s spectra, CeO2/mSiO2 after photodegradation of PFOS pollutants exhibit two characteristic peaks at 685.72 and 689.37 eV (Fig. 4E), which correspond to the F-C groups in organofluoride and the adsorbed F-Si groups, respectively, as confirmed by comparing reference samples before illumination and those soaked in NaF. Similarly, high-resolution XPS Si 2p spectra show an increasing peak at 104.81 eV after photodegradation or soaked in NaF solution (Fig. 4F, Supplementary Figs. S34–S36), attributed to Si-F bond formation. These results directly confirm that CeO2/mSiO2 catalyst can adsorb the free F− during the photodegradation of organofluorine. Besides, high-resolution XPS Ce 3d and O 1s remain at similar binding energies after photodegradation or NaF treatment, confirming the stability of catalysts. After the third cycle, the photodegradation ability of 5%CeO2/mSiO2 photocatalyst for OFB remains above 80% (Supplementary Fig. S37). This result also indicates that the photocatalytic activity primarily depends on the CeO2 component. The XPS S 2p spectra of PFOS (Supplementary Fig. S34C) and N 1s of FLE (Supplementary Fig. S36F) signals completely disappear after photodegradation, confirming organofluorine pollution was effectively mineralized. Furthermore, the mineralization of organofluorides by CeO2/mSiO2 catalyst is applicable to other organofluorides, such as PFOA and fluorinated e-waste DTMDEB (Supplementary Fig. S38).

Discussion

In summary, this work presents a potential technology for the removal of organofluorine pollutants by harnessing solar energy along a green and sustainable energy development path. CeO2/mSiO2 nanocomposites, synthesized via sol-gel and hydrothermal methods, possess a large specific surface area and the capability to absorb visible light. Fluorinated e-waste OFB, fluoro-antibiotic FLE, and perfluorinated pollutants PFOS were chosen as representative models featuring typical Csp2-F and Csp3-F bonds. Under visible light illumination, the optimal 5%CeO2/mSiO2 photocatalysts demonstrate efficient photodegradation performance for OFB (200 min, 99.5%), FLE (200 min, 94.4%) and PFOS (360 min, 98.3%). Notably, the degradation rate of PFOS by CeO2/mSiO2 reaches 25.9 ± 2.7% under outdoor sunlight within 5 days, significantly outperforming the parallel experiment with almost 0% degradation without catalyst. This marks the first report on the effective removal of ultra-stable PFOS under sunlight. Additionally, the influence of pH, anion and cation on the degradation performance of organofluorine pollutants was investigated through preliminary simulations of actual wastewater treatment processes. Regarding the degradation mechanism, DFT calculations illustrate the destruction pathways of Csp2-F and Csp3-F bonds by the generated •OH under illumination, which has a significant thermodynamic advantage over OH- under heating conditions. Notably, CeO2/mSiO2 can adsorb F ions, enabling the removal of organic pollutants with minimal fluorine emission, This study has achieved the solar-driven mineralization of fluorine-containing organic pollutants without F emission, offering a potential green, energy-saving, and sustainable technology for the treatment of fluorine-containing organic pollutants in wastewater management.

Methods

Chemicals and materials

Ce(NO3)3•6H2O (99.5%) was purchased from Aladdin, P. R. China. Tetraethyl orthosilicate (TEOS, 99%), Hexadecyl trimethyl ammonium bromide (CTAB, 99%), octafluoro-4,4’-biphenyldiamine (OFB, 98%) were purchased from Energy Chemical Co., Ltd. HCl, NaOH, Na2C2O4, isopropyl alcohol (IPA), 6,8-Difluoro-1-(2-fluoroethyl)-7-(4-methylpiperazin-1-yl)-4-oxo-1,4-dihydroquinoline-3-carboxylic acid (fleroxacin, FLE, 98%), 2-[difluoro-(3,4,5-trifluorophenoxy) Methyl]-5-(4-ethylphenyl)-1,3-difluoro-benzene (BTMDEB, 97%) were purchased from Bidepharm Medical Co., Ltd. (Shanghai, China). Perfluorooctanoic acid potassium salt (PFOS, 98%) and perfluorooctanoic acid (PFOA, 98%) were purchased from the J&K Scientific. Ammonium hydroxide (25%) was purchased from Mackin Co., Ltd. (Shanghai, China). All reagents used as received without any further purification.

Instruments

The morphology, elemental contents of the catalyst were characterized using Field Emission Scanning Electron Microscope and energy dispersive X-ray spectroscopy (SEM-EDS, Gemini 300, Germany) and transmission electron microscopy (TEM, JEM-F200, Japan). The crystalline structures of the catalysts were identified by X-ray powder diffraction (XRD, D8 ADVANCE, Germany). The surface area and pore size distribution were determined by the BET method from N2 adsorption-desorption isotherms at 77 K using ASAP 2020 PLUS HD88 (USA). X-ray photoelectron spectroscopy (XPS) was performed usedThermo ESCALAB 250Xi. Solid state UV–vis diffuse reflectance spectra were measured using a Lambda 950 UV/VIS Spectrometer equipped with an integrating sphere attachment, with BaSO4 power as background reference. Fourier transform infrared (FTIR) spectroscopy was performed over a range of 400~4000 cm−1 using a Nicolet iS50 spectrometer (USA). Electrochemical impedance spectroscopy (EIS) was collected using an electrochemical workstation (CHI660E, Shanghai Chenhua, China). Electron paramagnetic resonance (EPR, A300, BRUKER, USA) with 5, 5-dimethyl-1-pyrroline N-oxide (DMPO, J&K Scientific Ltd., Beijing) as radical capturer was used to detect the presence of hydroxyl radical (•OH) and superoxide radical (•O2−) under simulated sunlight irradiation. The concentrations of OFB, FLE, and DTMDEB were analyzed using a UV-8000 UV–vis analyzer (METASH, China), with detection wavelengths set at 267 nm, 270 nm, and 278 nm, respectively. TOC were measured on a SHIMADZU instrument (Total Organic Carbon Analyzer TOC-L CPH Basic System, TOC-L CPH).

Preparation and characterization of CeO2/mSiO2 nanocomposites

The mesoporous SiO2 (mSiO2) was synthesized by sol-gel method58. Firstly, 1.458 g of CTAB was dissolved in 60.0 mL deionized water, and 4.4 mL TEOS and 0.56 mL ammonia were added to trigger the reaction. Thus, the molar ratio of CTAB, TEOS, deionized water and ammonia was controlled as 0.01: 0.002: 1.6: 0.015. The mixed solution was stirred at room temperature for 5 h. As the reaction proceeded, the white turbid solution gradually transforms into a clear and transparent solution, and finally forms a viscous milky white sol. After filtration, the sol was washed with deionized water three times, dried at 80 °C for 12 h, and then ground into a powder sample using an agate grinding machine. Finally, the powder samples were calcinated in the muffle furnace at 550 °C for 6 h to obtain mSiO2.

The CeO2/mSiO2 with different ratios were synthesized using hydrothermal methods. Initially, varying amounts of Ce(NO3)3•6H2O were ultrasonically dispersed in 10 mL methanol, and then 0.95 g mSiO2 powders were added. The solvent was removed by vacuum rotary evaporation to ensure a complete mix Ce(NO3)3·6H2O and mSiO2. Subsequently, the mixed raw materials were transferred into the ceramic crucible and calcinated in the muffle furnace at 250 °C for 3 h to obtain CeO2/mSiO2 composites. A series of mSiO2 with varying CeO2 loadings were synthesized by adjusting the mass ratio of Ce(NO3)3•6H2O to mSiO2, which were named as 5%CeO2/mSiO2, 10%CeO2/mSiO2, 20%CeO2/mSiO2, 50%CeO2/mSiO2, and 80%CeO2/mSiO2, respectively.

Photodegradation experiments

Photodegradation experiments were conducted under the irradiations of a pure 410-nm LED lamp, and a 300 W xenon lamp as well as outdoor sunlight, respectively. Saturated aqueous solution of OFB (11.4 μmol L−1 at R.T.)46,48, 50.0 μmol L−1 FLE and 10.0 μmol L−1 PFOS were used as organofluorine pollutant solutions to evaluate their photodegradation performance. The light interval was set, and after each interval, 3 mL of the reaction solution was centrifuged and filtered to test the change in organofluorine concentrations. It is worth noting that the concentration of OFB and FLE were monitored by UV-Vis absorption spectroscopy. Whereas the concentrations of PFOS was detected by HPLC-MS. The degradation rate was calculated using the formula: (C0-Ct)/C0, where Ct is the concentration of degraded organo-fluorine pollutants, and C0 is the initial concentration of organofluorine pollutants. The optimization of the pH of the reaction solution was regulated by 0.1 M NaOH and HCl.

Intermediates and products analysis

The intermediates of OFB, PFOS and FLE were analyzed using a Thermo Ultimate 3000 Infinity HPLC System equipped with a Thermo TSQ ENDURA LC/MS System, which utilized electrospray ionization (ESI) in both positive and negative ion modes, along with Q1 Full Scan for the target analyzed. The chromatographic column used was an AC Agilent ZORBAX SB-C18 (2.1 mm × 100 mm, 1.8 μm). Prior to detection, all samples were filtered through 0.22 μm needle-like filters. The intermediate products were identified by liquid chromatography-mass spectrometry (LC-MS) analysis with the following solvent systems: for OFB, a mixture of water (20%) and acetonitrile (80%) at a flow rate of 0.3 mL min−1; for FLE, a mixture of methanol (25%) and water (75%) at a flow rate of 0.8 mL min−1; and for PFOS, 2 mM ammonium acetate in acetonitrile (100%) at a flow rate of 0.5 mL min−1. Mass spectrometry (MS) was performed in an electrospray ionization (ESI) mode, with the scanning mass ranged set from 50 to 500 m/z.

DFT calculations

DFT calculations for compounds and intermediates were performed using the Gaussian 09 program. For geometries optimizations and singlet point energy calculations, the M062X/6-31 g(d) and M062X/6-311 g(2 d) levels of theory were employed. Intrinsic reaction coordinate (IRC) calculations were executed to identify transition states and ensure their connection to the corresponding intermediates. The bond dissociation energies (BDE) were computed at an identical theoretical level59,60.