Introduction

As global energy demand grows and the need for a clean energy transition increases1, nuclear power is expected to be a promising new energy in the foreseeable future thus has been developed vigorously2,3. Uranium is the most important component to trigger the fission reaction4,5. Considering the limited uranium source in the land and the considerable amount of uranium resource in the sea (over 4 billion tons)6, enrichment of uranium from seawater is of great significance for the sustainable development of nuclear industrial7,8. However, due to the low concentration of uranyl in seawater (3-4 parts per billion) and the complicated marine environment9,10, developing an efficient enrichment approach for uranyl up-recycle remains a challenge.

Advanced porous materials such as Metal-Organic Frameworks (MOFs)11,12, Porous Organic Polymers (POPs)7,13 and Covalent-Organic Frameworks (COFs)14,15,16,17, have been extensively developed for uranium extraction from natural seawater, due to their functionalized building blocks and tunable pore characteristics18. The diverse functional groups and the well-designed adsorbents/photocatalysts ensured the high extraction capacity by virtue of the adsorption19,20 and/or photo-reduction of discrete uranyl at binding sites21,22 at enriched binding sites. For instance, Yang et al. designed a hierarchical porous membrane utilizing amidoxime-functionalized PIM-1, which allows rapid diffusion of uranium species and realized a remarkable 20-fold enhancement in uranyl adsorption capacity compared to the membrane with only intrinsic microporosity23. Liu et al. successfully anchored atomically dispersed Cu to a UiO-66-NH2 support via a simple ligand-assisted ice photocatalytic reduction pathway, achieving a high uranyl extraction capacity (9.16 ± 0.39 mg-U/g-Ads) from natural seawater24. In 2023, Yang et al. reported a series of novel intramolecular donor-acceptor (D-A) systems by introducing different electron-donating and withdrawing groups in the pores of the COF structure, which exhibit uranyl uptake capacity of 8.02 mg/g/day in real seawater under visible light25. However, due to the limited available binding sites within such framework material, it is hard to overtake the uptake capacity based on the extraction via the discrete uranyl-binding.

In our previous study26, the excited UO22+ can be utilized for photochemical production of hydrogen peroxide (H2O2), which can react with UO22+ to form solid studtite. Such photochemical process was applied for selective uranyl extraction from seawater without using any solid materials27. Nevertheless, due to the absence of nucleation center, the extraction must take a long time (at least 24 h). It was also reported by Kim et al. 28. that the Eq. (1) between H2O2 and uranyl at ambient condition are reversible.

$${{{\rm{UO}}}}_{2}^{2+}+{{{\rm{H}}}}_{2}{{{\rm{O}}}}_{2}+4{{{\rm{H}}}}_{2}{{\rm{O}}}\rightleftharpoons \left({{\rm{U}}}{{{\rm{O}}}}_{2}\right)\cdot \left({{{\rm{O}}}}_{2}\right)\cdot 4{{{\rm{H}}}}_{2}{{{\rm{O}}}}_{\left({{\rm{s}}}\right)}+2{{{\rm{H}}}}^{+}$$
(1)

When pH value is low enough (e.g., pH = 1), solid studtite can be efficiently dissolved as free UO22+. Inspired by these conclusions, we proposed to construct a series of photocatalysts with efficient H2O2 production activity and binding centers for studtite formation. More importantly, such COFs should be acid-resistant so that the possibly formed studtite can be dissolved under acidic condition and COFs can be regenerated simultaneously for long-term utilization.

H2O2 is a very famous quintessential green oxidant but has difficulty for long-term storage because of the self-decomposition29. Anthraquinone method is widely employed in industrial production of H2O2, contributing to over 95 % of H2O2 output30. The anthraquinone method involves the use of substantial organic solvents, generates toxic by-products, and is prone to catalyst deactivation, leading to high energy consumption and substantial pollution31. Although H2O2 is not combustible itself, its high oxidizing properties render it a hazardous industrial raw material. In particular, during closed transportation, high concentrations of H2O2 (above 74 %) can lead to a rapid temperature increase, potentially causing the gas phase to explode. Consequently, utilizing a green synthesis strategy to produce H2O2 for local utilization has become a research topic recently32,33. Since imine-based COFs are well-known photocatalysts for the synthesis of H2O234,35, we designed and synthesized four imine-based COFs photocatalysts (Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd) with chelating groups for uranyl such as keto groups and bipyridine N36,37,38,39 (Fig. 1). It is found that compared with COFs with individual keto or bipyridine N chelating group, the abundant binding sites on Tp-Bpy containing both keto groups and bipyridine N are beneficial for the formation of dissoluble studtite nanodots, which could realize the efficient studtite nanodots growth-elution cycle. To broaden the photocatalysts and clarify the suppose that combination of abundant adsorption sites and the H2O2 production ability of photocatalysts could lead to the highly efficient uranyl enrichment via studtite nanodots growth and elution cycle, we prepared another two imine-based COF photocatalysts with keto groups and pyridine-N sites (Tp-Bpy-2 and Tp-Py)40 (Supplementary Fig. 1,2), both of which showed efficient enrichment of uranyl in the studtite nanodots growth-elution cycles.

Fig. 1: Synthesis and structures.
figure 1

The schematic diagram for synthesis and structures of four COFs.

Full size image

Results

Synthesis and characterization of COFs

In this work, combination of diverse linkers was used to synthesize COFs with different functional groups, which endow the different capacity of H2O2 production. Initially, β-ketoenamine COFs were synthesized by altering the bipyridine monomer, obtaining Tp-Bpy and Tp-Bd via the condensation reaction in a mixture of N, N-dimethylacetamide (DMAC) and o-dichlorobenzene (o-DCB) with acetic acid as catalyst at 120 oC. Next, the COFs without any carbonyl groups were synthesized via the similar solvothermal reaction, which were assigned to Tb-Bpy and Tb-Bd. The structure of the four COFs was determined by powder X-ray diffraction (PXRD), solid 13C NMR Spectroscopy and Fourier infrared transform spectroscopy (FT-IR).

As clearly shown in Fig. 2a and Supplementary Fig. 3, the vanish of the N-H stretching band (-NH2) at 3200–3400 cm−1 and −CH = O stretches at 2800–3100 cm−1 confirmed complete consumption of the aldehyde and diamine groups in the reactants41,42. As for Tp-based COFs, the newly emerged C-N characteristic stretching peaks and the formed C = O and C = C bond at ~ 1585 cm−1 and ~ 1579 cm−1, respectively, verified the presence of keto form rather than enol form. As for Tb-based COFs, the C = N stretching at 1622 cm−1 perspicuously manifested the formation of imine bonds. Simultaneously, X-ray photoelectron spectrum (XPS) N 1 s distinctly designated three forms of N (C-N-, C = N and imine bonds) for the four COFs41,43 (Supplementary Fig. 2), which were in line with above-mentioned FT-IR results. Solid-state 13C CP/MAS NMR spectra equally confirmed the successful preparation of the COFs38,44. Tp-Bpy and Tp-Bd displayed distinctive signals for the carbonyl carbons (−C = O) at approximately 183 − 187 ppm and exocyclic carbons (−C = C) at approximately 107 ppm (Fig. 2b). On the other side, imine bonds (C = N) at ~ 157–162 ppm emerged in Tb-Bpy and Tb-Bd. The combination of FT-IR and 13C NMR analysis demonstrated the successful synthesis of four COFs at the chemical structure level.

Fig. 2: Chemical structure and characterization of Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd.
figure 2

a FT-IR spectra. b Solid-state 13C CP/MAS NMR spectra. c–f PXRD characterization (experimental PXRD profile: red, refined profile: thin blue, Bragg positions: green and difference: blue. The insets show the structural models and interlayer distances of each COF assuming an AA stacking mode. g–j SEM images.

Full size image

Furthermore, the crystallinity of each COF was examined by PXRD and Pawley refinement analysis. As shown in Figs. 2c–f, there are the strongest diffraction peaks in each PXRD pattern at 2θ = 3.93o, 3.55o, 3.88o and 3.63o, indicating that the open channel of the four COFs was formed successfully. Meanwhile, they all had a broad peak at around 2θ = 27o due to the π-π stacking between the COF layers corresponding to the (001) plane. Furthermore, the Pawley refinement of the experimental PXRD data was done to find out the details of the unit cell parameters (Supplementary Table 1–4), the detailed factors and negligible residuals (Rwp and Rp) of which were in accordance with the predictions. The characteristic of the surface morphologies of the four COFs was conducted utilizing Scanning Electron Microscopy (SEM). Figures 2g–j showed that Tp-Bpy and Tb-Bd exhibited blocky structure composed of aggregated microspheres, while Tp-Bd exhibited grass-like morphologies that consist of well-bedded nanofibers and Tb-Bpy showed a close aggregation of rod-like structures.

High specific surface area and adjustable porosity of COFs make them optimal photocatalysts with more exposed active sites. To acquire the porosity, N2 adsorption-desorption analyses were conducted on the fully activated samples at 77 K. The adsorption-desorption isotherms were identified as type-II curves with a mesoporous character14 (Fig. 3a). The total pore volumes were evaluated to be 0.90, 0.46, 0.50, and 0.33 cm3/g for Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd, respectively. The pore size distributions were calculated by BJH to be centered at 2.06 nm for Tp-Bpy, 2.09 nm for Tp-Bd, 2.24 nm for Tb-Bpy and Tb-Bd, respectively, which were all in good concordance with the calculation value (Supplementary Fig. 5). Furthermore, the Brunauer-Emmett-Teller (BET) surface areas of Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd are 1346.5, 972.4, 247.2 and 199.1 m2 g−1, respectively. The thermal stabilities of the four COFs were estimated by thermogravimetric analysis (TGA) under nitrogen atmosphere at 10 °C/min. The TGA curves unveiled that these COFs demonstrate exceptional thermal robustness, enduring temperatures as high as 410  °C almost without decomposition (Fig. 3b).

Fig. 3: Pore structure and photoelectrochemical tests for Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd.
figure 3

a N2 sorption isotherms. b Thermal gravimetric analysis spectra. c UV-vis diffuse reflection spectra. d Kubelka-Munk-transformed reflectance spectra. e Mott–Schottky curves. f Scheme illustration of the electronic band structures.

Full size image

Characterization of the four COFs in photoelectric properties

The visible-light-harvesting capacity and energy band structures of Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd were investigated by UV-Visible diffuse reflectance spectroscopy (UV-vis DRS) and Mott-Schottky (M-S) curves. As presented in Fig. 3c, all COFs exhibit intense absorption in the visible region. Analysis via Kubelka-Munk transformation equation in conjunction with Tauc plots (Fig. 3d) displayed that the energy band-gaps (Eg) of Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd were 1.70, 1.65, 2.23 and 2.32 eV, respectively. According to the M-S curves (Fig. 3e), it was obvious that all COFs were typical n-type semiconductors for a positive slope. Moreover, the flat band values (Efb, vs Ag/AgCl) were estimated to be −0.50, −0.49, −0.54 and −0.59 eV for Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd, respectively. The conduction band (CB) potential (ECB) is typically about 0.1 V lower than the flat band potential of n-type semiconductors45,46. Combining with the value of Eg and ECB, the valence band (VB) potential (EVB) could be plainly calculated (Fig. 3f). Furthermore, the energy level matched well with the requirements for the reduction of O2 to H2O2 (0.68 eV vs NHE)47,48, revealing that they are all promising photocatalysts for H2O2 production.

Photo-extraction uranyl investigations

To evaluate the photocatalytic performance, the photo-extraction of uranyl was tested at pH = 4 and 5. At pH = 4, as shown in Supplementary Fig. 6, approximately 88.8 % of uranyl can be extracted with Tp-Bpy, obviously higher than that with other three COFs. As for Tp-Bd, there was a stagnation of extraction to 58.0 % when the reaction time was up to three hours. Differently, we found that Tb-Bpy showed a relatively slow but steady extraction rate. At pH = 5, the overall reaction velocity had been significantly increased, with the final extraction efficiency of Tp-Bpy, Tp-Bd and Tb-Bpy increased to 96.3 %, 76.1 % and 91.0 %, respectively (Fig. 4a). Similarly, Tb-Bpy shows a steady uranyl extraction from 30–180 min, while the extraction rate on Tp-Bpy and Tp-Bd are slowed down obviously or even stagnated. With Tb-Bd or in the absence of catalysts, the removal of uranyl was negligible. Compared with the uranyl extraction under dark condition, it’s concluded that the efficient uranyl extraction over Tp-Bpy, Tp-Bd and Tb-Bpy is largely related with photo-reaction although the adsorption of uranyl over Tp-Bpy contributes about 48 % to the whole extraction efficiency (Fig. 4b and Supplementary Fig. 7). The adsorption of uranyl on Tb-Bd is the lowest, which may be related with the poor complexation ability for uranyl ions in the absence of chelating groups.

Fig. 4: Photocatalytic performance of Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd for uranyl extraction and H2O2 production.
figure 4

Reaction condition: mphotocatalyst = 10 mg, CUO22+ = 20 ppm, Cmethanol = 10 vol%, V = 40 mL, visible light, pH = 5, 25  °C and 3 h. Error bars are the standard deviations of three replicate measurements. a Photo-extraction kinetics. b Uranyl extraction efficiency under dark and light conditions. c H2O2 production kinetics in the presence of uranyl. d H2O2 production kinetics in the absence of uranyl.

Full size image

The quantities of H2O2 production were measured by colorimetric method (Fig. 4c, d and Supplementary Fig. 8, 9). The H2O2 production rate over Tp-Bpy, Tp-Bd, Tb-Bpy and Tb-Bd is about 123.89, 8.79, 23.48 and 11.17 μM/h, respectively. Without catalyst, there is no H2O2 produced in the 10 % methanol aqueous solution. Figure 3c showed the different trends of H2O2 production kinetics with UO22+, which was related with the complicated interplay of H2O2 production and consumption process with different photocatalysts. Firstly, without catalyst, uranyl itself could be excited under visible light to produce H2O226, with similar production rate as that for the Tb-Bpy (Fig. 4c). Secondly, the adsorption of uranyl on COFs could block the H2O2-production activity of both uranyl and COFs since the photogenerated electrons could not be directly transformed to O2 when the active sites such as keto or bipyridine N chelating group was occupied with uranyl47,49,50; Thirdly, H2O2 could be consumed via the reaction between H2O2 and uranyl. Based on the above processes, the different trends of H2O2 production can be explained. For the case of uranyl aqueous solution without catalyst, the detected H2O2 is from the H2O2 production of uranyl itself with no simultaneous consumption of H2O2 since uranyl is not transformed into studtite. For the case of Tb-Bd + uranyl, since uranyl was only ca. 10 % removed (Fig. 4a), H2O2 was almost not consumed to form studtite. The production of H2O2 from uranyl itself and Tb-Bd, accompanied with mutual intervention results in the observed trend. For the case of Tb-Bpy + uranyl, Tp-Bd + uranyl and Tp-Bpy + uranyl, the uranyl binding leaded to the obviously blocked H2O2 production, in which Tb-Bpy adsorbed the lowest uranyl, thus the initially detected H2O2 was a little higher than that with Tp-Bd and Tp-Bpy due to the minimal blocking from uranyl binding. With the uranyl extracted on COFs and the decreased concentration of uranyl in the solution, both H2O2 production and consumption were slower, thus there were somehow platform for the detected H2O2 in 30–60 min. When uranyl concentration was low enough, the production of H2O2 overpass the consumption of H2O2, the detected H2O2 amount goes up again. For the case of Tp-Bd, in Fig. 4d, it showed that the H2O2 production ability of Tp-Bd was quite poor and the adsorption of uranyl at the binding site seriously blocked the production of H2O2, thus the platform only emerged in 5–15 min, and there was almost no production of H2O2 after 60 min thus no uranyl could be extracted as studtite in Fig. 4a. Moreover, in Fig. 4a, Tb-Bpy exhibited a relatively slower extraction rate than Tp-Bd, but the production of H2O2 by Tb-Bpy was higher than that by Tp-Bd, which can be explained based on the following deduction. Within the first 90 minutes, although Tb-Bpy produced a higher quantity of H2O2 than Tp-Bd, Tp-Bd exhibited a significantly greater adsorption kinetics compared to Tb-Bpy as shown in Supplementary Fig. 7b. After 90 min, the rate of H2O2 production of Tp-Bd decreases due to the affection of adsorbed uranyl, leading to a stagnation in uranyl extraction. In contrast, Tb-Bpy continued to produce H2O2, thus uranyl concentration continues to decline after 90 min. In the earlier stage of extraction, the quick adsorption kinetics dominates the process, while in the lateral stage when the binding sites were occupied by uranyl, the uranyl adsorption was merely stopped and the reaction between uranyl and H2O2 played important role to the extraction. In brief, the different detected H2O2 trends shown in Fig. 4c was the results of complicated interplay of the production and consumption of H2O2 during the time course of uranyl extraction.

Combined analysis of uranyl extraction kinetics shown in Fig. 4a and H2O2 production shown in Fig. 3c, d and Supplementary Fig. 8, it suggests that i) the low H2O2 production rate and the poor adsorption for uranyl over Tb-Bd result in the negligible extraction of uranyl; ii) the limited H2O2 production rate after 90 min resulted in the stagnation of uranyl extraction; iii) the extraction kinetics over Tp-Bpy and Tb-Bpy is somehow different, i.e., the extraction rate decay over Tb-Bpy is slower than that over Tp-Bpy, which implies that the extraction paths may be different; iv) Tp-Bpy and Tb-Bpy are the promising photocatalysts for the efficient extraction of aqueous uranyl, although further recycling utilization has to be clarified.

Characterizations of photo-extraction products

The photo-extraction products obtained with Tp-Bpy and Tb-Bpy were further analyzed through XPS, FT-IR, PXRD, and High-resolution transmission electron microscopy (HRTEM). For both Tp-Bpy and Tb-Bpy, peaks at 382.20 eV and 381.95 eV can be assigned to the U 4f7/2 of U(VI)51,52 (Fig. 5a). As shown in Fig. 5b, a new peak at 918 cm−1 belonging to the bond U = O arisen for both Tp-Bpy and Tb-Bpy after photo-extraction53,54. Meanwhile, the unique carbonyl carbons (−C=O) and imine bonds (C=N) for respective Tp-Bpy and Tb-Bpy maintained well after photo-extraction. For Tb-Bpy, after photo-extraction, newly emerged diffraction peaks at 15o and 21o suggest the formation studtite particles, while for Tp-Bpy, no additional peaks can be observed (Fig. 5c). The WT maximum contour plots of Tp-Bpy after photo-extraction closely matched to the (UO2) (O2) ·4H2O plots, which strongly supported the formation of studtite nanodots (Fig. 5d). TEM image of Tb-Bpy after photo-extraction (Fig. 5e) shows obvious rodlike studtite nanoparticles on the surface. In contrast, no obvious nanoparticles can be found on the surface of Tp-Bpy (Fig. 5f). The HRTEM image (inset in Fig. 5f) showed two lattice fringe spacings of 0.265 and 0.325 nm, which can be ascribed to the (021) and (111) planes in studtite (PDF#49-1821). It suggests that the photo-extraction product on Tp-Bpy is studtite nanodots, while that on Tb-Bpy are much bigger studtite rodlike particles. Accordingly, the elemental mapping (Supplementary Fig. 10, 11) confirmed a homogenous distribution of U for each of them. For Tb-Bpy, U was distributed around the rodlike crystals, while for Tp-Bpy, U was dispersed on the whole surface without growth as big nanorods.

Fig. 5: Investigation of photo-extraction products for Tp-Bpy and Tb-Bpy.
figure 5

a XPS spectra after photo-extraction. b FT-IR spectrum comparison before and after photo-extraction. c PXRD after photo-extraction. d WT contour plots for (UO2) (O2)·4H2O and Tp-Bpy after photo-extraction. e HRTEM image of Tb-Bpy after photo-extraction. f HRTEM image of Tp-Bpy after photo-extraction.

Full size image

Growth-elution cycle for uranyl extraction

To recover the surface of Tp-Bpy and Tb-Bpy, the used photocatalysts were eluted with 0.1 M HNO3 for 1 h. The elution ratio of U from Tp-Bpy is about 90%, higher than that from Tb-Bpy (Supplementary Fig. 12). The obtained Tp-Bpy after elution was then used again for the photo-extraction of uranyl. After 7 consecutive extraction-elution cycles, more than 90% of extraction and elution efficiency for uranyl can be achieved (Supplementary Fig. 13). The FT-IR, SEM and XRD characterizations of the Tp-Bpy after extraction-elution cycling test suggest the high stability in the elution condition (Supplementary Fig. 14–16). In contrast, the consecutive extraction-elution cycles with Tb-Bpy are not satisfied, which may be attributed to the incomplete elution of bigger studtite particles. Notably, it is the formed studtite nanodots that makes the short-time elution feasible, which prevents the possible hydrolysis of imine-COFs in long-time elution. While the formed studtite nanorods cannot be completely eluted in short time before imine-COFs encounter hydrolysis.

To clarify the effect from co-existing metal ions, the photo-extraction of uranyl with Tp-Bpy was conducted in the presence of K+, Na+, Ca2+, Mg2+, Ba2+, Sr2+, HCrO4-, Cl- and CO32- with concentration 50 times as high as that of uranyl. Results showed that more than 80 % U(VI) can be extracted after 3 h irradiation, indicating that co-existing metal ions would not interfere the uranyl extraction (Supplementary Fig. 17).

Based on the analysis mentioned above, the four self-prepared COFs showed different performance: i) Tb-Bd has quite low extraction activity due to the low adsorption ability although H2O2 can be produced; ii) Tp-Bd shows unsatisfied extraction activity due to the limited H2O2 production despite of the considerable adsorption of uranyl. To confirm this, we supplemented the extra addition of H2O2 until the H2O2 concentration reaches up to 40 µmol/L and further conducted the uranyl extraction test with Tp-Bd (Supplementary Fig. 18). It turned out that the remained 34.8 % uranyl could further be extracted as studtite nanorods and only less than 6.2 % left in the solution after another 2 h reaction under dark condition; iii) Tb-Bpy shows a slowly decayed extraction kinetics with the formation of rodlike studtite particles that is difficult for elution, which is supposed due to the limited adsorption ability for uranyl; iv) the perfect extraction-elution cycling for uranyl enrichment was realized on Tp-Bpy. The four COFs are different with complexation groups, bipyridine and keto groups, which would also endow them with different photocatalytic performance for H2O2 production. It has been identified that the reaction between H2O2 and uranyl has a relatively low kinetics, which means the production of H2O2 on Tp-Bpy and Tb-Bpy is not the rate-determined-step55. Therefore, the different performance between Tp-Bpy and Tb-Bpy is supposed related with the nucleation of studtite on the surface of COFs, which is largely dependent on the chelating site for uranyl (Fig. 6).

Fig. 6: The schematic diagram for the reaction process of Tp-Bpy and Tb-Bpy.
figure 6

The abundant chelating sites for uranyl on Tp-Bpy are beneficial for the formation of dissolvable studtite nanodots. The limited adsorption ability for uranyl on Tb-Bpy is beneficial for the formation of large, hard-to-dissolve studtite nanorods.

Full size image

To broaden the photocatalysts and clarify the suppose that combination of abundant chelating sites and the H2O2 production ability of photocatalysts could lead to the highly efficient uranyl enrichment via studtite nanodots growth and elution cycle, we prepared another two imine-based COF photocatalysts with keto groups and pyridine-N chelating sites (Tp-Bpy-2 and Tp-Py). The successful preparation of their chemical structures was demonstrated by PXRD (Supplementary Fig. 19), FTIR (Supplementary Fig. 20), 13C NMR (Supplementary Fig. 21) and XPS (Supplementary Fig. 22). The possibility of H2O2 production was verified by a series of physical and photoelectric characterizations (Supplementary Fig. 23–25). Similar efficient tendency of uranyl adsorption (Supplementary Fig. 26), H2O2 production (Supplementary Fig. 27, 28), studtite nanodots growth-elution cycle (Supplementary Fig. 29–33) were found with Tp-Py and Tp-Bpy-2 as photocatalyst. The excellent growth-elution cycles of Tp-Bpy-2 and Tp-Py revalidated that multiple uranyl chelating sites promoted the growth of studtite nanodots and facilitated the efficient extraction of uranyl (Supplementary Fig. 34–37).

In summary, the requisites for the formation of dissolvable studtite nanodots include the presence of chelating sites for uranyl and the photosynthetic ability for H2O2, as well as the stability in elution condition. Herein, more than introducing a series of COFs with highly efficient extraction capacity, we developed a strategy for cyclable generation and elution of studtite nanodots for long-term uranyl extraction.

Enrichment of uranyl in natural seawater

In order to verify whether the COFs with multiple chelating sites has the potential to be applied for uranyl extraction from real seawater, uranyl with low concentrations (10 ppm and 1 ppm) in spiked seawater was used for photo-extraction with Tp-Bpy as photocatalyst, in which above 90 % of uranyl can be extracted (Fig. 7a). In addition, 94 % of 1 ppm uranyl in seawater can be extracted even in the presence of variable valency metal ions (Fig. 7b). After five consecutive extraction-elution cycles with 200 mL spiked seawater (1 ppm uranyl), 4 mL of concentrated solution with 28.7 ppm uranyl was obtained (Fig. 7c). To further investigate the practical applicability of the mentioned photo-extraction process, we examined the efficiency of U(VI) extraction from natural seawater under visible light. After calculating the prime cost for the preparation of the COFs, we choose the cost-efficient Tp-Py photocatalyst for the extraction and enrichment of uranyl from seawater in scale-up test (Supplementary Table 5 and 6). Owing to the complex marine environment of microorganisms, the higher antimicrobial ability against the marine microorganism community must be considered for efficient uranyl extraction in natural seawater56,57. The antifouling activity of Tp-Py is tested by using marine bacteria as targets (Fig. 7d). The result showed under dark condition (Supplementary Table 7), 24.8 % of the marine microbial community from natural seawater can be inhibited. After exposure to visible light, a markedly increased inhibition of 96.4 % was realized, which is supposed due to the production of reactive oxygen species (ROS) in the presence of visible light irradiation, as confirmed by the electron paramagnetic resonance (EPR) spectra (Fig. 7e and Supplementary Fig. 38). The experiments with natural seawater (3.3 ppb) demonstrated that Tp-Py achieved an impressive uranyl uptake capacity of 154.50 mg/g over 12 consecutive days under visible light, averaging 12.875 mg/g/day (Fig. 7f, Supplementary Fig. 39 and Supplementary Table 8, 9).

Fig. 7: Uranium extraction from natural seawater.
figure 7

a Removal rate at 10 and 1 ppm spiked seawater over Tp-Bpy. b Selectivity in 1 ppm uranyl spiked seawater over Tp-Bpy. c Diagram of enrichment with Tp-Bpy. d The antibacterial activity of Tp-Py against marine bacteria. The exact sample size (n): Top- nBlank = 472, nDark = 355. Down- nBlank = 417, nDark = 15. The value of n is the standard deviations of three replicate measurements. e EPR spectra for ·O2--DMPO, ·OH-DMPO, 1O2-TEMP complexes generated by Tp-Py under visible light irradiation. f Equipment used for uranyl extraction from natural seawater by Tp-Py.

Full size image

Discussion

In summary, we reported an efficient and sustainable strategy to upcycle the uranyl resource from natural seawater with COF photocatalysts. Comprehensive comparison study of different COFs Tp-Bpy, Tp-Py, Tp-Bpy-2 and other three COFs (Tp-Bd, Tb-Bpy, and Tb-Bd) photocatalysts on the adsorption capacity, the H2O2 production ability, the uranyl extraction-elution cycling performance showed that i) the photogenerated H2O2 can react with uranyl to form studtite; ii) the chelating site for uranyl on the photocatalyst is necessary for nucleation of studtite; iii) the ultra-highly efficient enrichment of uranyl can be realized via the formation of dissolvable studtite nanodots based on the abundant chelating sites for uranyl, which avoided the formation of large, hard-to dissolve studtite nanorod. After five consecutive extraction-elution cycles, 200 mL of 1 ppm uranyl-spiked seawater was concentrated to 4 mL of 28.7 ppm uranyl solution with regenerable Tp-Bpy photocatalyst. Furthermore, an impressive uranyl uptake capacity of 12.875 mg/g/day can be achieved with cost-efficient Tp-Py in real seawater under visible light irradiation. It’s believed that based on our results, more economical and acid-resistant COFs or even other materials with the necessary requisites, i.e., abundant uranyl-chelating sites and H2O2 production ability, could be prepared in the field of material engineering for the practical uranyl extraction from seawater.

Methods

Materials

All reagents were purchased from commercial suppliers and used without further purification unless stated otherwise. 1,3,5-Triformylphloroglucinol (Tp, 98 %), 1,3,5-Triformylbenzene (Tb, 97 %), 2,2′-bipyridine-5,5′-diamine (Bpy, 97 %), and Benzidine (Bd, 97 %) were purchased from Jilin Chinese Academy of Sciences-Yanshen Technology Co., Ltd. N, N-dimethylacetamide (DMAC, 99.8 %), 1,3,5-Trimethylbenzenewas (97 %, AR), o-dichlorobenzene (o-DCB, 98 %), 1-Butanol (99 %, AR), 1,4-Dioxane (99.5 %), acetic acid (AcOH, ≥ 99.8 %), ethanol absolute (99.5 %, Water ≤ 300 ppm) and Tetrahydrofuran (THF, 99 %, AR) were all provided by Aladdin. Ultrapure water was prepared from the Millipore system (18.25  MΩ·cm).

Synthesis of Tp-Bpy and Tp-Bd

1,3,5-Triformylphloroglucinol (Tp, 63 mg, 0.3 mmol) and 2,2’-bipyridine-5,5’-diamine (Bpy, 83.7 mg, 0.45 mmol) or Benzidine (Bd, 82.9 mg, 0.45 mmol) were put into a 15 mL Schlenk tube. Then a mixed solvent of N, N-dimethylacetamide (DMAC) and dichlorobenzene (3/1, v/v, 6 mL) was also added, and subjected to ultrasonic treatment to achieve uniform dispersion. Then 0.6 mL of 6 M acetic acid (AcOH) aqueous solution was quickly added and sonicated again. After degassing, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, water and THF and dried under vacuum at 80 °C for 12 h.

Synthesis of Tb-Bpy

1,3,5-Triformylbenzene (Tb, 32.43 mg, 0.2 mmol) and 2, 2’-bipyridine-5, 5’-diamine (Bpy, 55.8 mg, 0.3 mmol) were put into a 15 mL Schlenk tube. Then a mixed solvent of o-dichlorobenzene and Butanol (n-BuOH) (1/1, v/v, 3 mL) were also added, and subjected to ultrasonic treatment to achieve uniform dispersion. Then 0.3 mL of 6 M acetic acid (AcOH) aqueous solution was quickly added and sonicated again. After sonicating for 10 min, the tube was then flash frozen via liquid N2 bath and degassed by three freeze-pump-thaw cycles. Then the tube was sealed off and heated at 120 °C for 3 days. The solid was washed away with a large amount of ethanol and dried under vacuum at 80 °C for 12 h.

Synthesis of Tb-Bd

1,3,5-Triformylbenzene (Tb, 29 mg, 0.18 mmol) and Benzidine (Bd, 50 mg, 0.27 mmol) were put into a 15 mL Schlenk tube. Then a mixed solvent of 1,4-Dioxane and Butanol (1/1, v/v, 3 ml) was added, and subjected to ultrasonic treatment to achieve uniform dispersion. Then 0.3 mL of 6 M acetic acid (AcOH) aqueous solution was quickly added. After sonicating for 10 min, the tube was flash frozen via liquid N2 bath and degassed by three freeze-pump-thaw cycles. Then the tube was sealed off and heated at 120 °C for 3 days. The solid was washed away with a large amount of ethanol and dried under vacuum at 80 °C for 12 h.

Synthesis of Tp-Py and Tp-Bpy-2

1,3,5-Triformylphloroglucinol (Tp, 63 mg, 0.3 mmol) and 2,5-Diaminopyridine (Py, 49.11 mg, 0.45 mmol) or [3,3’-Bipyridine]−6,6’-diamine (Bpy-2, 70.28 mg, 0.45 mmol) were put into a 15 mL Schlenk tube. Then a mixed solvent of N, N-dimethylacetamide (DMAC) and dichlorobenzene (3/1, v/v, 6 mL) was also added, and subjected to ultrasonic treatment to achieve uniform dispersion. Then 0.6 mL of 6 M acetic acid (AcOH) aqueous solution was quickly added and sonicated again. After degassing, the reaction was conducted at 120 °C for 3 days. The solid was then washed away with a large amount of DMAC, water and THF and dried under vacuum at 80 °C for 12 h.

Characterizations

Powder X-ray diffraction (PXRD) analyses were performed using a Rigaku SmartLab SE X-ray diffractometer equipped with a Cu Kα source (small angle X-ray scattering data collected on a Bruker D8 Advance diffractometer was used to correct the deviation) with a step size of 0.01°. Fourier transform infrared spectra (FT-IR) were recorded on a SHIMADZU IRTracer-100. BET surface areas were determined from N2 adsorption/desorption isotherms collected at 77 K using Micromeritics TriStar II. Scanning electron microscopy (SEM) images were recorded on a Hitachi SU 8100 Scanning Electron Microscope. Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) images were recorded on a JEOL transmission electron microscope (JEM-F200) operating at an accelerating voltage of 200 kV. Solid-state 13C CP/MAS NMR spectra were collected on a BRUKER AVANCE NEO 400WB spectrometer. X-ray photoelectron spectroscopy (XPS) analyses were performed using a Thermo Scientific ESCALAB 250Xi spectrometer, equipped with a monochromatic Al Kα X-ray source (1486.8 eV). Photoelectrochemical experiments measurements were performed on an electrochemical workstation (CHI760E, CHI Instruments, Shanghai, China). Electron paramagnetic resonance (EPR) spectra were recorded at 293 K with a Bruker EMXnano259 spectrometer, operated at 9.62 GHz with 12.59 mW power and modulation at 100 kHz/1 G. UV-vis spectroscopy results were recorded in diffuse reflectance (DR) mode at room temperature on a SHIMADZU UV-2700 spectrophotometer equipped with an integrating sphere attachment. Thermogravimetric analyses (TGA) were carried out on a NETZSCH STA 2500 instrument. Inductively coupled plasma mass spectrometry (ICP-MS) analyses were performed on an Agilent 7800 spectrometer.

Photoelectrochemical measurements

The Mott-Schottky plots, photocurrent response and electrochemical impedance of the photo-catalysts were measured on an electrochemical workstation (CHI760E, CHI Instruments, Shanghai, China). A 300 W Xe lamp was utilized as the light source and Na2SO4 (0.5 M, pH = 6.18) aqueous solution was used as the supporting electrolyte throughout the photocurrent measurements. A platinum wire and Ag/AgCl electrode were used as counter electrode and reference electrode, respectively. 1 μL naphthol was dissolved in about 3 mL ethanol with 1 mg catalyst. Then, 50 μL was added to ITO with a layer of adhesive tape on the edge for six times, followed by drying in air. The potential vs NHE was determined by using the following Eq. (2):

$${{\rm{E}}}\left({{\rm{VS}}}.\,{{\rm{NHE}}}\right)={{\rm{E}}}\left({{\rm{VS}}}.\,{{\rm{Ag}}}/{{\rm{AgCl}}}\right)+0.197\, {{\rm{V}}}$$
(2)

Uranyl extraction experiments

All experiments on the extraction of U(VI) were carried out in air under visible light (400<λ < 800 nm) using 10 W LED lamp (Perfect PCX50C Discover) as light source and the maximum 100% photocurrent was corresponding to 10 W.

Photo-extraction of aqueous U(VI)

10 mg photocatalyst was dispersed in 40 mL of a uranyl spiked seawater (20, 10, 1 ppm) and ultra-pure water (20, 10 ppm), which contains 10 vol% methanol. The temperature was kept at 25 oC by circulating cooling water throughout this photo-extraction process. After photo-extraction, the resulted suspension was filtrated by 220 nm membrane. The concentration of UO22+ was measured by UV–vis spectrophotometry at a wavelength of 650 nm using the Arsenazo III method.

Photo-extraction uranyl from natural seawater

10 mg Tp-Py was dispersed in self-assembly designed reactor. The seawater flowed through the photocatalyst at the speed of 2.1 L h−1 by using peristaltic pump. The concentration of uranyl in the natural seawater was analyzed by inductively coupled plasma-Mass Spectrometry (ICP-MS).

Design of the device for uranyl extraction from natural seawater

10 mg COFs was mixed with 7 mL ethanol, 1 mL ultra-pure water and 20 μL 5 % of Nafion solution. The mixture was drop-casted onto quartz panel. With the peristaltic pump, the real seawater flowed through the coated quartz panel.

Enrichment Test

The used photocatalyst with uranyl uploaded was put into 1 mL of elution solution at room temperature. One hour later, it was taken out for another photo-extraction in uranyl-containing object solution for 3 h. The elution operation was repeated in the same elution solution to enrich uranyl. After 5 times of elution, the uranyl concentration of the final elution solution was detected by ICP-MS.

Effect of competing ions on the photo-extraction of aqueous U(VI)

10 mg photocatalyst was dispersed in 40 mL of ultra-pure water (20 ppm UO22+) with 1000 mg/L M(NO3)x·n(H2O) (M = K+, Na+, Ca2+, Mg2+, Ba2+, Sr2+, HCrO4-) or sodium salts (Cl- and CO32-) for 3 h irradiation.

Recyclability test

After one run of photo-extraction, the photocatalyst was regenerated by elution with HNO3 solution (pH = 1 or 2.5). The resulted photocatalyst was then reused for another photo-extraction experiment. For 10 mg catalysts, 40 mL elution solution was used to elute the binding uranium for 60 min at room temperature. The elution efficiency (E, %) was determined by using the following Eq. (3):

$${{\rm{E}}}=\frac{{{{\rm{C}}}}_{{{\rm{e}}}}\times {{{\rm{V}}}}_{{{\rm{e}}}}}{\left({{{\rm{C}}}}_{0}-{{{\rm{C}}}}_{{{\rm{t}}}}\right)\times {{{\rm{V}}}}_{{{\rm{t}}}}}\times 100\%$$
(3)

where Ce (mg L−1) is the uranium concentration in elution solution, Ve (L) is the volume elution solution, Ct (mg L−1) is the uranium concentration in uranium-spiked seawater after photo-reaction, Co (mg L−1) is the initial uranium concentration of uranium-spiked seawater, Vt (L) is the volume of uranium-spiked seawater used for adsorption. The resulting suspension was filtered and washed with ultra-pure water till the supernatant became neutral. After being dried under vacuum, the resultant material was used for another photo-reaction experiment. It was found that after six consecutive cycles Tp-Bpy still showed excellent uranium uptake.

H2O2 detection methods

The H2O2 generation was qualified by using a colorimetric method. Typically, the mixture after photo-reaction was centrifuged and then 500 μL of the suspension was added to a mixture of 2 mL KI solution (0.1 M) and 0.05 mL (NH4)2MoO4 solution (0.01 M). Corresponding absorption at 352 nm was measured on a UV-Vis spectrometry.

Calculation of the equilibrium constant for the Eq. (1)

42.5 mL of 50 ppm U(VI) solution was mixed with 7.5 mL MeOH in a 100 mL reactor. The reactor was thermostated at 25 °C with a cycling pump. The pH of the mixture was adjusted to 4 by adding negligible volumes of NaOH and HNO3 solutions (~ 1.0 M). The mixed solution was illuminated under visible light (λ > 420 nm) using a 300 W Xe lamp (Microsolar 300, PerfectLight) equipped with a UV cutoff filter, during which, uranyl species could be excited to produce H2O2 and the produced H2O2 then reacts with uranyl to form studtite solid. The concentration of uranyl in the solution was monitored every one hour in order to ascertain the equilibrium. It was found after 8 h, the concentration of remaining uranyl stopped decreasing. Subsequently, the final suspension stirs in the dark for an hour. Then the concentration of H2O2 and pH value were measured. The equilibrium constant for the reaction can be calculated with the detected concentrations of uranyl and H2O2, as well as pH value.

Anti-biofouling measurements

The marine bacterial strain was used to test the antimicrobial activity of the catalyst. The Tp-Py was dispersed into marine bacteria/fresh Luria-Bertani broth at a dispersion concentration of 0.25 mg/mL. After cultivated at 37 °C for 3 h with thermostatic oscillator, the viability of the bacterium was determined. The dilution plate counting method was used according to the Chinese standard GB/T20944 to determine the viability of the adsorbents treated bacterial cultures. In order to simulate the experimental conditions, the antibacterial effect was compared respectively under dark state and visible light irradiation. Inhibition rates (IR) were calculated according to the following Eq. (4):

$${{\rm{IR}}}=\frac{\left({{{\rm{C}}}}_{{{\rm{i}}}}-{{{\rm{C}}}}_{{{\rm{a}}}}\right)}{{{{\rm{C}}}}_{{{\rm{i}}}}}\times 100\%$$
(4)

Where Ci (CFU/mL) and Ca (CFU/mL) are the microbial concentrations in cultures before and after treatment with Tp-Py.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.