Electrochemical upcycling of uranyl from radioactive organic wastewater with a self-standing covalent-organic framework electrode

Characterization of PAE-COF-AO@CC electrode

As schematically illustrated in Fig. 1, we develop the in situ growth of a polyarylether-based COF (PAE-COF-CN) on the carbon cloth to construct a self-standing PAE-COF-CN@CC electrode material, which was then subjected to amidoximation to give PAE-COF-AO@CC. The scanning electron microscopy (SEM) images of the carbon cloth, PAE-COF-CN@CC, and PAE-COF-AO@CC (Fig. 2a–c) show that the sheet structures were uniformly grown on the smooth surface of the carbon cloth to form the brown flake material PAE-COF-CN@CC. After amidoximation, the morphology of PAE-COF-AO@CC was similar to that of PAE-COF-CN@CC. Comparison of X-ray diffraction (XRD) patterns of the carbon cloth, PAE-COF-CN@CC and PAE-COF-AO@CC implied that the characteristic sharp diffraction peak at 4.27° corresponding to (100) facet of PAE-COF-CN and PAE-COF-AO³⁷ (Fig. 2d and Supplementary Tables 1, 2), demonstrating that PAE-COF-CN and PAE-COF-AO were successfully loaded onto the carbon cloth substrate. The X-ray photoelectron spectrum (XPS) of O 1s and C 1s shown in Supplementary Fig. 1 confirmed the existence of abundant C=O and C–OH groups on the surface of carbon cloth. Thus, it is anticipated that the intimate interaction between PAF-COF-AO and carbon cloth is possibly enabled by the presence of oxygen-containing functional groups on carbon cloth, which facilitate the formation of effective hydrogen bonds between tetrafluoroterephthalonitrile (TFPN) and C-OH groups or the grafting of TFPN by nucleophilic aromatic substitution reactions (SN_Ar) with C-OH groups.

The schematic diagram for synthesis and structures of PAE-COF-CN@CC and PAE-COF-AO@CC.

a–c SEM images of PAE-COF-CN@CC, PAE-COF-AO@CC. Insets show the morphology of PAE-COF-CN@CC and PAE-COF-AO@CC electrodes at a higher magnification. d XRD patterns of PAE-COF-CN@CC, PAE-COF-AO@CC. The blue region represents the (100) facet of PAE-COF-CN and PAE-COF-AO. e FT-IR spectra of PAE-COF-CN@CC, PAE-COF-AO@CC. The blue region represents the peak of C≡N in PAE-COF-CN. f The solid-state ¹³C NMR spectra of PAE-COF-AO@CC.

Fourier transfer infrared (FT-IR) spectra in Supplementary Fig. 2 shows that compared with the HHTP and TFPN monomer, there are two new-emerged peaks at 1260 and 1024 cm⁻¹ ascribed respectively to dioxin C-O asymmetric and symmetric stretching vibrations, indicating the successful synthesis of PAE-COF-CN loaded on carbon cloth substrates^38,39. The FT-IR spectrum of PAE-COF-AO@CC showed that the peak at 2240 cm⁻¹ disappeared utterly compared with PAE-COF-CN@CC, whereas new peaks at 1651 cm⁻¹, 913 cm⁻¹ and 3171 cm⁻¹ emerged, which are assigned to the stretching vibration of C=N, N-O, and O-H, confirming the successful introduction of amidoxime groups³⁷ (Fig. 2e). The peaks at 139 ppm and 143 ppm in the solid-state ¹³C NMR spectra of PAE-COF-AO@CC (Fig. 2f) could be assigned to the carbon atoms in the phenyl group, confirming the formation of the COF skeleton³⁸. Meanwhile, the appearing chemical shift at 164 ppm suggests the presence of amidoxime groups on PAE-COF-AO@CC.

Electrochemical Upcycling of uranyl with the self-standing PAE-COF-AO@CC electrode

As shown in Supplementary Fig. 3, compared with pure carbon cloth electrode, the extraction performance from PAE-COF-AO@CC is remarkably promoted. The effect of pH on the electrolytic uranyl extraction over the PAE-COF-AO@CC electrode was investigated. As shown in Fig. 3a, when pH is as low as 3.0, only 30% of uranyl is extracted. At pH 4.0, the extraction sharply rises to 91.41%, and the optimal extraction of 97.96% is obtained at pH 5.0. Even in an alkaline matrix with a pH from 6 to 13, the uranyl extraction efficiency still exceeded 90%. For the further test, pH value of the electrolyte was set at 5.0. With the applied voltage increased from −0.193 to −0.493 V (vs. Ag/AgCl), the electrochemical uranyl extraction of the PAE-COF-AO@CC rose steadily from 62.35 to 97.96% (Fig. 3b). At different U/Na ratios, the uranyl extraction is all higher than 85% (Fig. 3c). For all the tested electrolyte with uranyl concentrated from 10 to 50 ppm, the electrochemical extraction was much more efficient compared with physical adsorption, and the promotion became even greater in the case of higher uranyl concentration (Fig. 3d–f). An extraction of 99.02% and an extraction capacity of 660.16 mg/g were achieved with 50 ppm of uranyl. The continuous utilization of such PAE-COF-AO@CC for electrochemical uranyl extraction showed the slowly decayed electro-activity (Fig. 3h) and that the extraction efficiency can be kept as high as ca. 90% after 204 h (Supplementary Fig. 4). It is found that the yellow deposits were kept on the electrode without falling into the electrolyte during the long-time utilization, while the color is changed from dark to light yellow (Fig. 3g and Supplementary Fig. 5).

The electrochemical extraction efficiency of uranyl using PAE-COF-AO@CC electrode a under different pH conditions (2–13), b at different potentials (−0.193 to −0.493 V (vs. Ag/AgCl)), c with different ion strength. Comparison of uranyl extraction with the electrochemical method and the physical adsorption with various uranyl concentrations, d ~10 ppm, e ~30 ppm, f ~50 ppm. g Visual change of PAE-COF-AO@CC electrode. h Chronoamperometry curve of PAE-COF-AO@CC at −0.493 V (vs. Ag/AgCl). The error bars in a–f denote the standard deviation where n = 3.

Mechanism of electrochemical extraction of uranyl

Comparison of cyclic voltammetry (CV) curves of PAE-COF-AO@CC electrode (Fig. 4a) in air-saturated and Ar-saturated NaNO₃ electrolyte showed that a characteristic oxygen reduction peak at −0.016 V (vs. Ag/AgCl) in the presence of O₂. The simultaneous production of H₂O₂ is confirmed by analysis of the electrolyte with colorimetric method (Supplementary Figs. 6,  7a). As reported, the aromatic ether groups are active for the electrocatalytic production of H₂O₂^40,41. As shown in Fig. 4b and Supplementary Fig. 8, with the addition of 100 ppm uranyl, the oxygen reduction peak at −0.016 V (vs. Ag/AgCl) was remained. With the continuous electrochemical extraction of uranyl, this oxygen reduction peak is gradually decreased. The extraction performance with an Ar-saturated electrolyte is remarkably lower than that with an air-saturated electrolyte (Supplementary Fig. 7b). All these results suggest that the uranyl extraction is related to the electrocatalytic production of H₂O₂. XRD pattern of the resulting electrode (Fig. 4c) suggests that the extracted uranyl on the electrode surface was identified as (UO₂)O₂·4H₂O species (PDF #16-0206), which is also referred to as studtite. To thoroughly clarify the extraction mechanism, the precipitate at different time intervals was collected for XRD, XPS, in situ Raman, and SEM analysis. As indicated by XRD patterns, the precipitate at 1, 3, and 5 h were ascribed to studtite, and the diffraction peaks from studtite became more obvious as the reaction time increased (Supplementary Fig. 9). The corresponding XPS analysis implied that the immobilized U on the electrode was U(VI) rather than U(IV) (Fig. 4d–f). The in situ Raman spectra in Supplementary Fig. 10a showed the emerged and gradually increased two peaks at 824 and 865 cm⁻¹ assigned to studtite within the initial 3660 s^42,43,44, which is in line with the characteristic peaks from the final precipitates on the electrode obtained in our main set-up after 5 hours-electrochemical reaction (Supplementary Fig. 10b). The peak at ~481 cm⁻¹ is assigned to the adsorbed UO₂²⁺ on the surface^44,45. Notably, there is no Raman signal from U(IV) in the initial stage of electrochemical reaction according to the obtained in situ Raman spectra, suggesting that the adsorbed uranyl directly reacts with H₂O₂ to form studtite rather than reduced to U(IV) intermediate^44,46. SEM image of the used electrode after 1-h-reaction showed the presence of small nanoparticles sized around 20 nm (Supplementary Fig. 11). All these characterizations suggest that the uranyl extraction is attributed to the growth of studtite, without any intermediate U(IV) species. With further extraction test, the particles grew gradually and finally formed a dense layer on the electrode surface (Fig. 4g and Supplementary Figs. 12, 13). The energy spectrum and element mapping images confirmed the presence of U and O (Supplementary Fig. 14). The H₂O₂ production activity of the used PAE-COF-AO@CC electrode was compared as shown in Supplementary Fig. 15. The electrode after 24 h extraction showed similar activity as the fresh electrode, while electrode extracted for 168 h or longer showed seriously decayed performance. Thus it is anticipated that in the initial growth of studtite, the H₂O₂ production is not interfered. It is no doubt that with the growth of studtite, the active site for H₂O₂ production would be blocked.

a CV curves of PAE-COF-AO@CC electrode in 2 mM NaNO₃ electrolyte saturated with Ar and air at a scan rate 50 mV/s; b CV curves of the PAE-COF-AO@CC electrode in 2 mM NaNO₃ electrolyte saturated with air in the presence of uranyl at different time intervals at a scan rate 50 mV/s. c XRD patterns of the used PAE-COF-AO@CC electrode. U 4f XPS spectrum of PAE-COF-AO@CC electrode after d 1 h, e 3 h, and f 5 h uranyl extraction. g SEM images of PAE-COF-AO@CC electrode after various uranyl extraction times. Insets show the morphology of the electrodeposited particles after various uranyl extraction times at a higher magnification. h EXAFS R-space and corresponding fitting curves for PAE-COF-AO@CC after uranyl adsorption. i Comparison of uranyl adsorption capacity of the used PAE-COF-AO@CC electrode. The error bars denote the standard deviation where n = 3.

Compared with the extraction performance over the PAE-COF-CN@CC electrode, the uranyl extraction over the PAE-COF-AO@CC electrode is about thrice boosted (Supplementary Fig. 16a, b), which is supposed due to the different chelating ability of the terminal cyano and amidoxime groups. In the FT-IR spectra of PAE-COF-CN@CC and PAE-COF-AO@CC before and after uranyl adsorption, the peak of PAE-COF-AO@CC at 871 cm⁻¹ became broad and stronger after uranyl loading (Supplementary Fig. 16c), which is likely ascribed to superposition of the new U-O stretching vibration and the red-shifted N-O stretching vibration⁴⁷. In comparison, FT-IR spectra of PAE-COF-CN@CC after uranyl loading remain unaltered (Supplementary Fig. 16d), suggesting that amidoxime groups had stronger coordination interaction with uranyl than cyano groups.

The Fourier transform (FT) k³-weighted extended X-ray absorption fine structure (EXAFS) was performed to probe the uranyl coordination environment in PAE-COF-AO@CC. The FT curve showed peaks at 1.40 and 1.90 Å, which could readily be assigned to a U-O_ax and a mixture of U-O_eq + U-N scattering paths, respectively. Fitting the spectrum confirmed that in PAE-COF-AO@CC, uranyl is bonded in a uranyl-amidoxime η²-binding mode^48,49, where a uranyl center is coordinated by two axis oxygen atoms, two atoms from coordinated H₂O molecules, and oxygen atoms and nitrogen atoms from amidoxime groups (Fig. 4h, Supplementary Fig. 17, and Supplementary Table 3). It’s worth noting that the chelating of uranyl onto the amidoxime groups can work even in the presence of large quantities of organic matter, e.g., 800 ppm of polyvinyl alcohol, 12,000 ppm of urea, 8000 ppm of tetrahydrofurfuryl alcohol.

The adsorption of uranyl over the used PAE-COF-AO@CC electrode was shown in Fig. 4i. It is clear that even covered by studtite, uranyl adsorption on the electrode could be kept quite well, which is supposed due to the chemical bonds formed between uranyl and the surface O^2- or -OH groups^15,44,50. Although the detected amount of H₂O₂ decayed gradually, there is still enough H₂O₂ produced for extraction of uranyl via the formation of studtite since the reaction to form studtite is slow compared with the production of H₂O₂.

Based on the above, the efficient electrochemical uranyl extraction by the indirect electrochemical method is attributed to the following (Fig. 5): (i) the strong chelating ability of amidoxime groups toward uranyl ions that ensured the nucleation of studtite; (ii) the continuous production of H₂O₂ and adsorption of uranyl of the covered PAE-COF-AO@CC that ensured the lasting growth of studtite until the dense layer formed and the H₂O₂ production completely stopped.

PAE-COF-AO@CC provided chelating sites for uranyl ions and nucleation center for the following growth of studtite from the reaction between H₂O₂ and uranyl.

It has been shown in Fig. 3a that at pH ≤3.0, the extraction performance is remarkably decayed. Such a phenomenon can be explained by two factors: (1) based on Eq.(1), the formation of studtite will be inhibited under acidic conditions; (2) at low pH, amidoxime groups undergo protonation, leading to electrostatic repulsion with uranyl species as well as the sluggish nucleation and growth of studtite.

Extraction performance in low-level radioactive organic wastewater

We further applied such cost-efficient self-standing COF electrodes for uranyl extraction from low-level uranyl-containing radioactive organic wastewater (Supplementary Table 4). Figure 6a showed that after 24 h with a constant voltage of −0.493 V (vs. Ag/AgCl), an extraction efficiency of 93.80% and an extraction capacity of 592.94 mg/g were achieved, which was six times higher compared to that with physical adsorption (Fig. 6a). Continuous extraction upon 450 h could lead to an accumulated extraction capacity of 9238.9 mg/g (Fig. 6b). XRD pattern of the resulted PAE-COF-AO@CC electrode confirmed the formation of (UO₂)O₂·4H₂O (Fig. 6c). Most importantly, PAE-COF-AO@CC continued to exhibit excellent uranyl extraction capacity at pH 9, which matches well with the real low-level uranyl-containing radioactive organic wastewater^8,51. Supplementary Fig. 18 showed that after 24 h with a constant voltage of −0.730 V (vs. Ag/AgCl) at pH 9, an extraction efficiency of 93.09% and an extraction capacity of 620.58 mg/g were achieved. The satisfied growth of studtite in the complex organic matrix suggested that the chelating of uranyl and nucleation of studtite are not interfered by the co-exist organics as revealed by EXAFS analysis (Fig. 6d, Supplementary Fig. 19, and Supplementary Table 5). The resulting PAE-COF-AO@CC electrode loaded with studtite after 480 h-electrochemical processes was further calcined in air at 800 °C. XRD analysis demonstrated that after calcination, high-pure U₃O₈ was obtained (Supplementary Fig. 20). Economic evaluation considering materials and operation cost showed that with such an indirect electrochemical method, the extraction cost is around 0.50 USD per gram of U (Supplementary Table 6).

a Comparison of uranyl extraction from low-level uranyl-containing radioactive organic wastewater with the electrochemical method and the physical adsorption. Error bars are the standard deviations of three replicate measurements. b Long-term uranyl extraction performance from low-level uranyl-containing radioactive organic wastewater. The error bars denote the standard deviation where n = 3. c XRD patterns of PAE-COF-AO@CC electrode after uranyl extraction from low-level uranyl-containing radioactive organic wastewater. d FT-EXAFS spectra and corresponding R-space fitting curves of PAE-COF-AO@CC loaded with uranyl from stock uranyl solution and low-level uranyl-containing radioactive organic wastewater.