Hollow Ga-Nx-C capsule synthesis and characterizationThe synthesis of the hollow Ga-Nx-C capsules is shown in Fig. 2a. In brief, ZIF-8 and ZIF-8@K-TA nanocrystals were synthesized following a previously reported process with slight modification31,32. Then, ZIF-8@K-TA was added into a methanol solution containing gallium ions to exchange the potassium ions, yielding ZIF-8@Ga-TA which retained the crystalline texture of the parent ZIF-8 nanocrystals, as confirmed by powder X-ray diffraction (PXRD) and Fourier transform infrared (FT-IR) spectroscopy (Supplementary Figs. 1 and 2). Scanning electron microscopy (SEM) revealed ZIF-8@Ga-TA retained the characteristic polyhedral morphology of ZIF-8 (Supplementary Fig. 3). Subsequently, hollow Ga-Nx-C capsules were obtained by pyrolysis of ZIF-8@Ga-TA under an argon atmosphere.Fig. 2: The synthesis and structural characterization of Ga-Nx-C.a Schematic illustration of the synthesis of Ga-Nx-C. b, c SEM and TEM images of Ga-Nx-C. d HAADF-STEM images and corresponding EDS maps revealing a homogeneous distribution of C (green), N (yellow), and Ga (blue) in Ga-Nx-C. e Aberration-corrected HAADF-STEM image of Ga-Nx-C, showing the atomically dispersed gallium. f N 1s XPS spectrum for Ga-Nx-C. g Ga K-edge XANES spectra (inset: expanded view showing the Ga valence states). h FT k3-weighted χ(k) EXAFS spectra. i EXAFS fitting curves for Ga-Nx-C. j N2 adsorption-desorption isotherms for Ga-Nx-C.SEM and transmission electron microscopy (TEM) images revealed that the Ga-Nx-C capsules consisted of very thin carbon layers (Fig. 2b, c). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and corresponding energy dispersive spectroscopy (EDS) elemental mapping images showed that C, N, and Ga were distributed uniformly in Ga-Nx-C (Fig. 2d). Moreover, atomically dispersed gallium sites were evidenced in the aberration-corrected HAADF-STEM images (Fig. 2e), which showed bright spots of atomic size. X-ray photoelectron spectroscopy (XPS) analysis of Ga-Nx-C verified the presence of C, N, and Ga in the sample (Supplementary Fig. 4). The N 1s spectrum was deconvoluted into pyridinic N (398.6 eV), pyrrolic N (400.7 eV), graphitic N (401.7 eV), and oxidized-N (404.7 eV) species (Fig. 2f)28,33.Next, X-ray absorption near-edge structure (XANES) and Fourier-transformed extended X-ray absorption fine structure (EXAFS) measurements were conducted on Ga-Nx-C to determine the valence state and the local coordination environment of the atomically dispersed gallium species. Ga foil, Ga2O3, and gallium(III)-(4-methoxycarbonylphenyl) porphyrin (Ga-TCPP) were used as reference samples34. As shown in Ga K-edge XANES spectra, the absorption edge position for Ga-Nx-C was located between the Ga foil and Ga2O3, suggesting the valence state of Ga in Ga-Nx-C was between 0 and +3 (Fig. 2g). Furthermore, the absorption edge of Ga-Nx-C was close to that of Ga-TCPP, suggesting the valence state of Ga in Ga-Nx-C was likely between +2 and +3. Possibly, some of the Ga(III) ions in ZIF-8@Ga-TA were reduced to Ga(II) during high-temperature pyrolysis step used to synthesize Ga-Nx-C. EXAFS showed that Ga-Nx-C and Ga-TCPP both had peaks at ~1.46 Å, which could readily be assigned to a Ga-N scattering path (Fig. 2h). Both pyridinic N and pyrrolic N can bind gallium ions, forming porphyrin-like GaN4 sites on the capsules. The peak at ~2.7 Å for Ga-TCPP was attributed to a metal-carbon (i.e., Ga-C) second shell of the porphyrinic ring34. The wavelet transform (WT) EXAFS plot of Ga-Nx-C showed the maximum at R(~1.46 Å)/k(~4.38 Å), which could be assigned to Ga-N bonding (Supplementary Fig. 5). No obvious peak due to Ga-Ga bonding was observed, indicating the absence of gallium nanoparticles. The fitting results confirmed a Ga-N coordination number ~4, consistent with the porphyrin-like GaN4 structure (Fig. 2i and Supplementary Table 2). The porosity of Ga-Nx-C was determined by N2 sorption at 77 K. The adsorption-desorption isotherms were type IV with an obvious adsorption/desorption hysteresis at higher P/P0 values, suggesting a hierarchical porous structure with abundant mesopores (Fig. 2j). The calculated Brunauer-Emmett-Teller (BET) surface area of Ga-Nx-C was 972.80 m2 g−1. A pore size distribution analysis using the nonlocal density functional theory (NLDFT) method afforded an average pore size centered at 15 nm (Supplementary Fig. 6). Ga-Nx-C contained 1.29 wt.% gallium, as determined by inductively coupled plasma mass spectrometry (ICP-MS). Porous N-doped carbons supporting atomically dispersed gallium catalysts have been shown previously to offer excellent electrocatalytic performance in reactions such as the carbon dioxide reduction reaction35. Therefore, we hypothesized that the developed Ga-Nx-C capsules should be efficient catalysts for the electrocatalytic reduction/extraction of uranium from aqueous solution.Electrochemical uranium extraction performanceUranium extraction from mining waste, nuclear waste, seawater, and contaminated water remains an attractive proposition for both nuclear fuel supply and environmental remediation16,36,37,38,39. Electrochemical uranium extraction methods are regarded as a promising alternative to traditional chemical processing methods for U extraction and recycling. Although enormous efforts have been focused on this aspect, until now there have been no reports of the electrochemical generation of U(IV) solids such as UO2 under ambient conditions. In this light, we set about exploring the electrochemical uranium extraction properties of Ga-Nx-C under various conditions.We first tested the U(VI) redox properties of Ga-Nx-C capsules by running cyclic voltammetry (CV) in uranium-spiked groundwater (~500 ppm). Ga-Nx-C loaded on carbon felt was used as the working electrode. Ag/AgCl and graphite rod were used as reference and counter electrodes, respectively. A reduction peak of U(VI) is observed at a potential (V vs. Ag/AgCl) of −0.41 V, indicating that U(VI) was reduced to U(IV) (Fig. 3a). No U(VI) reduction peaks were observed in the absence of Ga-Nx-C (Fig. 3b). Subsequently, we studied the uranium extraction performance of Ga-Nx-C using a square wave conversion method in ~1.2 ppm and ~120 ppb uranium-spiked groundwater. The uranium concentration was reduced from ~1.2 ppm (and ~120 ppb) to lower than 13 ppb (and 3 ppb) in 1440 min, suggesting a high removal efficiency (Fig. 3c, d). The extraction capacities of Ga-Nx-C were calculated to be 22.63 and 2.36 mg g−1, respectively. Notably, 27.6% and 38.2% calcium removal were detected in ~1.2 ppm and ~120 ppb uranium-spiked groundwater, respectively. The magnesium concentration was reduced from ~5.1 ppm to 2.1 ppm (removal percentage of 58.9%) and 2.254 ppm (removal percentage of 55.8%) under the same conditions. To further investigate the generation products through electrocatalysis, we performed extraction experiments in ~20 ppm uranium-spiked groundwater. As shown in Fig. 3e, some pale yellow products formed on the working electrode during electrocatalysis in ~20 ppm uranium-spiked groundwater. These pale yellow products were collected and examined by PXRD, SEM, and XPS, revealing the generation of CaCO3 nanosheets and Mg(OH)2 nanosheets attached to the Ga-Nx-C capsules (Fig. 3f, g and Supplementary Fig. 7). Besides Na, K, Zn, Mg, and Ga metals, the XPS spectrum showed the presence of uranium after electrocatalysis (Supplementary Fig. 7). The U 4f XPS spectrum of the pale yellow products showed U 4f7/2 and U 4f5/2 peaks at 381.6 and 392.3 eV, respectively, corresponding to a U(IV) species (Fig. 3h)15. No UO2 signals were observed by PXRD, which might be due to the small size of the nanoparticles. However, we observed peaks due to UO2 nanocrystals in the PXRD pattern of the pale yellow product from ~50 ppm uranium-spiked groundwater (Supplementary Fig. 8). We further carried out high-resolution TEM (HRTEM) measurements to analyze the components in the pale yellow products. HRTEM images of some of the larger particles revealed lattice fringes with interplanar distances of around 3.0 Å and 2.3 Å, which were assigned to the (104) planes of CaCO3 and (101) planes of Mg(OH)2, respectively (Supplementary Fig. 9). Moreover, some dark-color nanoparticles with a diameter of ~6–7 nm were observed, showing lattice spacings of 3.2 Å, which can be assigned to the (111) plane of face-centered cubic UO2 nanoparticles (Fig. 3i)15. No solid products were obtained when performing experiments in uranium-spiked deionized water under similar conditions (Supplementary Fig. 10). These results indicate that U(VI) was reduced to U(IV) in the presence of Ca(II), Mg(II), or other metal ions in the groundwater, which appeared to promote the formation of solid UO2.Fig. 3: Electrochemical uranium removal performance of Ga-Nx-C in uranium-spiked groundwater and product analysis.a, b Cyclic voltammograms of Ga-Nx-C/carbon felt and carbon felt (blank) in uranium-spiked groundwater (~500 ppm). c, d Uranium extraction from spiked groundwater with initial uranium concentrations of ~1.2 ppm and ~120 ppb using Ga-Nx-C. e Photograph showing the generated pale yellow product formed through electrocatalysis in ~20 ppm uranium-spiked groundwater (inset: expanded view showing the working electrode). f SEM image of the pale yellow product formed through electrocatalysis in ~20 ppm uranium-spiked groundwater. g PXRD for the pale yellow product formed through electrocatalysis in ~20 ppm uranium-spiked groundwater. h U 4f XPS spectrum of the pale yellow product formed through electrocatalysis in ~20 ppm uranium-spiked groundwater. i HRTEM images of UO2 nanoparticles generated through electrocatalysis in ~20 ppm uranium-spiked groundwater (inset: expanded view showing the lattice spacing.).The above results inspired us to further investigate the effects of secondary metal ions on uranium extraction from aqueous solutions. Next, electrocatalytic extraction of uranium experiments were conducted using deionized water containing uranium and one type of secondary metal ion. Initially, CV tests were conducted on Ga-Nx-C to study the reduction voltage of U(VI) in aqueous solution in the presence of each different type of metal ion. The obtained U(VI) reduction voltages were used for uranium extraction studies in the presence of the different metal ions (Supplementary Fig. 11). In the presence of Na+, K+, or Cs+ ions, no solid was obtained after 24 h electrocatalysis, suggesting that U(VI) was not reduced to U(IV) during catalysis (Fig. 4a and Supplementary Fig. 12). XANES revealed that the valence state of the uranium adsorbed on the Ga-Nx-C electrode surface was +6 in all these solutions (Fig. 4b)15,40. EXAFS and corresponding fitting results matched data for UO2(NO3)2·6H2O, suggesting no U(IV) product was generated in aqueous solutions containing alkali metal ions (Fig. 4c and Supplementary Fig. 13).Fig. 4: Product analysis for the electrochemical reduction of U(VI) by Ga-Nx-C in solutions containing selected secondary metal ions.a Photograph showing no solids were generated through electrocatalysis on Ga-Nx-C in ~100 ppm uranium-spiked sodium nitrate solution. b, c U L3-edge XANES and EXAFS spectra of Ga-Nx-C after electrocatalytic uranium extraction in the presence of Na+, K+, or Cs+ ions (b inset: expanded view showing the U valence states). d PXRD pattern of the electrocatalytically generated products in ~100 ppm uranium-spiked cadmium nitrate solution. e U L3-edge XANES spectra of Ga-Nx-C after electrocatalytic uranium extraction in the presence of Mg2+, Ca2+, Sr2+, Ba2+, Cd2+, or Pb2+ ions (inset: expanded view showing the U valence states). f U L3-edge EXAFS curves and fitting results for Ga-Nx-C after uranium extraction in ~100 ppm uranium-spiked cadmium nitrate solution. g PXRD pattern of electrocatalytically generated Ce0.8U0.2O2. h HRTEM image of electrocatalytically generated Ce0.8U0.2O2 (inset: expanded view showing the lattice spacing). i U L3-edge EXAFS curves and fitting results for electrocatalytically generated Ce0.8U0.2O2. The reference data for UO2 and UO2(NO3)2·6H2O in (b, c) were taken from our previous work28.We further conducted experiments in Mg2+, Ca2+, Sr2+, Ba2+, Cd2+ and Pb2+ aqueous solutions containing uranium. Pale yellow solids were obtained after electrocatalysis for all these ions (Supplementary Fig. 14). PXRD showed that Mg(OH)2, CaCO3, SrCO3, BaCO3, CdCO3, and PbCO3 were obtained, respectively (Fig. 4d and Supplementary Fig. 15). No uranium-containing crystalline solid was detected in the presence of Mg2+, Ca2+, Sr2+, or Ba2+, indicating the formation of amorphous or nanosized uranium-containing particles. UO2 peaks were detected in uranium-spiked cadmium nitrate and uranium-spiked lead nitrate solutions (Fig. 4d). XANES and EXAFS spectra confirmed the existence of UO2 in these products, conclusively demonstrating that U(VI) could be electrocatalytically reduced to U(IV)O2 in the presence of these metal ions (Fig. 4e, f and Supplementary Fig. 16). Notably, the UO2 peaks were detected by PXRD for experiments conducted in the presence of Cd2+ or Pb2+ ions, further supporting these conclusions (Fig. 4d and Supplementary Fig. 15).Next, the electrocatalytic experiments were carried out in aqueous Ni2+, Zn2+, Al3+, and Fe3+ solutions. Metallic nickel, zinc oxide, Fe-containing and Al-containing amorphous solids were obtained after electrocatalytic reduction, as revealed by PXRD analysis (Supplementary Fig. 17). The formation of UO2 species was verified by XANES and EXAFS spectra (Supplementary Fig. 18).We next studied the effects of lanthanide and actinide metal ions such as Ce4+, La3+, Eu3+, and Th4+ ions. Interestingly, a crystalline bimetallic oxide phase, Ce0.8U0.2O2, was generated in the presence of Ce4+ ions41. The PXRD showed peaks at 28.8°, 33.3°, 47.7°, 56.6°, and 76.9°, corresponding to (111), (200), (220), (311), and (331) reflections of Ce0.8U0.2O2 (Fig. 4g). The HRTEM image of the Ce0.8U0.2O2 product showed lattice fringes with an interplanar spacing of 3.1 Å, which matched the contrast profiles of the (111) planes (Fig. 4h). To further probe the electronic structure of the Ce0.8U0.2O2 product, XANES and EXAFS measurements were performed. The U L3-edge XANES spectrum of Ce0.8U0.2O2 was similar to that of the UO2 standard sample, with the edge position being typical for U(IV) (Supplementary Fig. 19). Further, the EXAFS spectrum exhibited main peaks at 1.04 Å and 1.42 Å, corresponding to the first and second U-O coordination shells in Ce0.8U0.2O2, respectively (Supplementary Fig. 19). Fitting the EXAFS data to the Ce0.8U0.2O2 structural model was successful (fitting results indicated that the U atoms were coordinated by four O atoms, Fig. 4i). When experiments were carried out in aqueous La3+, Eu3+, or Th4+ solutions, amorphous powder products were obtained, evidenced by PXRD measurements (Supplementary Fig. 20). XANES and EXAFS analysis revealed a similar U(IV) structural model to that of the Ce0.8U0.2O2 (Supplementary Fig. 19). Annealing of amorphous samples is a well-known method to synthesize larger crystallites, thereby giving sharper and more intense PXRD peaks. Therefore, we further annealed each amorphous sample at 800 °C in a nitrogen atmosphere to verify the presence of bimetallic metal oxides. As expected, the PXRD patterns for the annealed samples revealed the presence of crystalline Eu0.76U0.43O2, La0.9U0.3O2, and Th0.75U0.25O2 (Supplementary Fig. 20)41,42,43.In-situ Raman spectroscopy was next applied to gain deeper understanding of the electrocatalytic processes leading to the generation of Ce0.8U0.2O2 in Ce4+/UO22+ aqueous solutions. K+/UO22+ aqueous solution was used for comparison. Before square wave potential cycling, U(VI) signals was detected at 870 cm−1 that could be assigned to adsorbed UO22+ ions (Fig. 5a)44,45. As the reaction time increased, new signals appeared at 257 cm−1 and 743 cm−1 suggesting the generation of U(IV)44,45. No U(IV) signals were observed in the presence of K+ (Fig. 5b). These results indicated that the U(VI) was reduced to U(IV) in the presence of Ce4+ ions (and by analogy La3+, Eu3+, and Th4+ ions) leading to the generation of bimetallic metal oxides, consistent with the X-ray absorption spectroscopy (XAS) and PXRD results.Fig. 5: Electrochemical removal of uranium by Ga-Nx-C in a simulated LMW solution.a In situ Raman spectra collected from a Ga-Nx-C/carbon felt working electrode in ~100 ppm uranium-spiked cerous nitrate solution. b In situ Raman spectra collected from the Ga-Nx-C/carbon felt working electrode in ~100 ppm uranium-spiked potassium nitrate solution. c Uranium extraction under LMW conditions, using Ga-Nx-C as an electrocatalyst. d Concentration changes of various ions before and after uranium extraction under LMW conditions, using Ga-Nx-C as the electrocatalyst. e PXRD pattern of the electrocatalytically generated products under LMW conditions. f SEM image of Ga-Nx-C after uranium extraction under LMW conditions. g Ga 2p XPS spectrum after uranium extraction under LMW conditions. h Large-scale extraction of uranium from spiked groundwater by the Ga-Nx-C electrocatalyst. i Concentration changes of various ions before and after uranium extraction from spiked groundwater, using Ga-Nx-C as the electrocatalyst.On the basis of these findings, we then performed electrocatalytic uranium extraction studies using Ga-Nx-C in a simulated low and media level radioactive waste (LMW) solution46,47. As shown in Fig. 5c, Ga-Nx-C capsules could rapidly remove uranium in the simulated LMW solution, achieving a removal capacity of 1015.9 mg g−1 in 48 h. Notably, the Ce4+ was also quickly removed as the UO22+ was consumed. The concentrations of other metal ions in the LMW solution were barely affected by electrocatalysis (Fig. 5d). The PXRD pattern of the generated powder showed peaks matching Ce0.8U0.2O2, suggesting that the reduced U(IV) has an exceptional binding affinity towards Ce4+ relative to other ions (Fig. 5e). The SEM image showed that Ga-Nx-C retained its hollow capsule morphology, indicating good structure stability (Fig. 5f and Supplementary Fig. 21). XPS revealed that Ga, N, and C were retained on the surface of Ga-Nx-C after catalysis, further confirming its good stability (Fig. 5g and Supplementary Fig. 22). We further investigated the electrochemical durability of Ga-Nx-C in a large-scale extraction experiment using 50 L of uranium spiked groundwater (~1 ppm). A uranium extraction capacity of 2.23 g g−1 was achieved after 108 h, suggesting Ga-Nx-C would be a very promising electrocatalyst for uranium extraction from contaminated groundwater (Fig. 5h, i).Electrochemical mechanism studiesOn the basis of the above findings, we next explored mechanism of U(IV) oxide (including UO2 and bimetallic MxUyO2, M = Ce, Th, La, Eu) formation using molecular dynamics (MD) simulations48. Since electrocatalysis generates the UO2 in the presence of alkaline earth metal ions and transition metal ions, whilst yielding MxUyO2 in the presence of lanthanide and actinide metal ions, we first studied the dynamic interactions between U(IV) and selected representative secondary metal ions (such as K+, Ca2+, Fe3+, or Ce4+) in aqueous solutions. Figure 6a shows the snapshots from the MD simulations in the presence of K+, NO3-, and H2O molecules. Initially, K+ and U(IV) were uniformly dispersed in the aqueous solution. No aggregation of K+ and U(IV) were observed after 100 ps or 500 ps, suggesting a very weak interaction between K+ and U(IV). The corresponding radial distribution functions (RDFs) through the MD simulation revealed the interactions between U(IV) and NO3− or H2O molecules were much stronger than with K+ (Fig. 6b). Moreover, mean square displacement (MSD) analysis showed that the diffusion coefficients of K+ and U(IV) differed significantly, thus allowing re-oxidation of the generated U(IV) to U(VI). Thus, no solid U(IV) products formed (Fig. 6c). This helps to explain why no UO2 or MxUyO2 were generated in the presence of alkali metal ions. In contrast, U(IV) and Ca2+ aggregated after 100 ps, and the aggregation intensified after 500 ps (Fig. 6d). Compared to K+, RDFs showed similar diffusion coefficients for U(IV) and Ca2+, as well as stronger interactions, which likely facilitated the co-generation of UO2 and CaCO3 precipitates during electrocatalysis (Fig. 6e, f). Furthermore, similar results were obtained in the presence of Fe3+, with even stronger interactions between U(IV) and Fe3+ (Fig. 6g–i). Notably, U(IV) and Ce4+ aggregated within a very short period of time, with the emergence of a very sharp and intense peak at r ~3.7 Å by RDFs reflecting a very strong interaction between the two ions (Fig. 6j–l). Figure 6c, f, i, l and Supplementary Fig. 23 show the corresponding diffusion coefficient plots and relative concentration distribution analysis for the U(IV)/K+, U(IV)/Ca2+, U(IV)/Fe3+, and U(IV)/Ce4+ systems. The diffusion coefficient gap and relative concentration difference between U(IV) and K+ gradually increased, while very similar diffusion coefficients and relative concentrations were seen for U(IV)/Ca2+, U(IV)/Fe3+, and U(IV)/Ce4+ during the electrocatalysis process. Again, the data suggested that Ca2+, Fe3+ and Ce4+ ions promote the generation of UO2 or UxMyO2 in the aqueous solutions. The MD simulations provide theoretical justification for the production of UO2 and UxMyO2 solids in the presence of certain secondary metal ions.Fig. 6: Mechanism analysis of secondary metal ion-induced electrochemical U(VI) reduction to U(IV) solid products.a Selected snapshots from the MD simulation showing the K+/U(IV) interactions on Ga-Nx-C in ~100 ppm uranium-spiked potassium nitrate solution. b RDFs plots of U(IV)-K+, U(IV)-O, U(IV)-O(H2O), and U(IV)-O(NO3−), and their corresponding average coordination number. c MSD-time curves for the transport of U(IV)/K+ (showing the different diffusion coefficients). d Selected snapshots from the MD simulation showing the Ca2+/U(IV) interactions on Ga-Nx-C in ~100 ppm uranium-spiked calcium chloride solution. e RDFs plots of U(IV)-Ca2+, U(IV)-O, U(IV)-O(H2O), and U(IV)-O(NO3−), and their corresponding average coordination number. f MSD-time curves for the transport of U(IV)/Ca2+. g Selected snapshots from the MD simulation showing the Fe3+/U(IV) interactions on Ga-Nx-C in ~100 ppm uranium-spiked ferric nitrate solution. h RDFs plots of U(IV)-Fe3+, U(IV)-O, U(IV)-O(H2O), and U(IV)-O(NO3−), and their corresponding average coordination number. i MSD-time curves for the transport of U(IV)/Fe3+. j Selected snapshots from the MD simulation showing the Ce4+/U(IV) interactions on Ga-Nx-C in ~100 ppm uranium-spiked cerous nitrate solution. k RDFs plots of U(IV)-Ce4+, U(IV)-O, U(IV)-O(H2O), and U(IV)-O(NO3−), and their corresponding average coordination number. l MSD-time curves for the transport of U(IV)/Ce4+.
