Highly selective urea electrooxidation coupled with efficient hydrogen evolution

Materials characterizationThe atomically isolated asymmetric Ni–O–Ti sites were constructed by anchoring single Ni atoms onto a commercial monolithic Ti foam. The key to the synthesis of asymmetric Ni–O–Ti sites is the in situ generation of coordinated unsaturated Ti with dangling bonds on the amorphous TiOx surface layer of the Ti foam under a reductive 5.0% H2/Ar atmosphere (Supplementary Figs. 1, 2)30. These unsaturated Ti atoms could provide reliable immobilization sites for single Ni atoms via strong metal-support interactions. Aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) revealed that the Ni atoms were atomically dispersed on the surface of the TiOx (Fig. 2a), which was highlighted by the corresponding line and 3D surface intensity profile aligned with the STEM simulation (Fig. 2b–d). Complementary STEM-mapping and X-ray diffraction (XRD) revealed the uniform dispersion of Ni atoms without discernible Ni clusters or nanoparticles (Fig. 2e and Supplementary Fig. 3). Inductively coupled plasma‒mass spectrometry (ICP-MS) analysis revealed a Ni loading of ~ 0.29 wt.% (Supplementary Table 1). For comparison, the connected symmetric Ni–O–Ni sites on the Ti foam anode were synthesized by increasing the Ni loading to 1.08 wt.% (Supplementary Fig. 4).Fig. 2: Characterization of the atomically isolated asymmetric Ni–O–Ti sites.a Atomic-resolution HAADF-STEM image of asymmetric Ni–O–Ti sites with white circles marking Ni atoms. b Corresponding line intensity profile taken along the white dashed rectangle, and (c) 3D surface intensity profile in the yellow dashed square box. d 2D black and white STEM simulation images on the basis of the Z contrasts of atomic columns and corresponding structure diagram of asymmetric Ni–O–Ti sites. e STEM elemental mapping. f Ni K-edge X-ray absorption near-edge structure spectra of the atomically isolated asymmetric Ni–O–Ti sites and connected symmetric Ni–O–Ni sites on Ti foam. g FT-EXAFS spectra and (h) wavelet transformed (WT) k3-weighted χ(k)-function of Ni K-edge of the asymmetric Ni–O–Ti sites and symmetric Ni–O–Ni sites. Ni foil and NiO were used as the standard samples.We then employed X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure spectroscopy (EXAFS) to decipher the chemical state and coordination environment of atomically isolated asymmetric Ni–O–Ti sites. The absorption edge energy of Ni in the asymmetric Ni–O–Ti sites was between those of standard Ni foil and NiO, suggesting the positively charged nature of the Ni atoms (Fig. 2f). In contrast, the absorption edge energy of connected symmetric Ni–O–Ni sites on Ti foam was similar to that on Ni foil, indicating of a more metallic state and consistent with the X-ray photoelectron spectroscopy (XPS) results (Supplementary Fig. 5). The Fourier-transformed EXAFS spectra of the asymmetric Ni–O–Ti sites displayed a dominant peak at 1.59 Å and a minor peak at 2.69 Å (Fig. 2g), which were attributed to Ni–O coordination in the first shell and Ni–Ti coordination in the second shell, respectively. The k-space of the minor peak located at 6.1 Å–1 differed from the 7.0 Å–1 k-space of the Ni–Ni coordination found in Ni foil, NiO, Ni(OH)2, or symmetric Ni–O–Ni sites according to the wavelet-transform EXAFS (Fig. 2h and Supplementary Fig. 6). Notably, the peak for metallic Ni–Ni scattering at 2.09 Å in the asymmetric Ni–O–Ti sites was absent, highlighting the atomically dispersed nature of the Ni atoms. Quantitative structural analysis indicated that each Ni atom in the asymmetric Ni–O–Ti sites was coordinated with approximately five adjacent Ti atoms in the second shell, which were interconnected by five O atoms (Supplementary Figs. 7, 8 and Table 2). As a comparison, the Ni atoms within the Ni–O–Ni sites on the Ti foam and conventional Ni-based materials were exclusively coordinated with Ni to form connected symmetric Ni–O–Ni sites.Electrochemical performancesAccording to linear sweep voltammetry (LSV), the atomically isolated asymmetric Ni–O–Ti sites exhibited low potentials of 1.30 V and 1.33 V vs. reversible hydrogen electrode (VRHE) to achieve laboratory- and industrial-level current densities of 10 mA cm–2 and 100 mA cm–2, respectively, while the connected symmetric Ni–O–Ni counterparts required slightly higher potentials (Fig. 3a and Supplementary Figs. 9–12). For comparison, freestanding Ti foam was ineffective at catalyzing the UOR, while the Ni foam showed poor UOR activity, demanding considerably higher potentials of 1.38 and 1.54 VRHE to reach equivalent current densities of 10 mA cm–2 and 100 mA cm–2, respectively. Notably, the Ni–O–Ti sites had a substantially smaller Tafel slope (14.2 mV dec–1) than did the Ni–O–Ni sites (20.0 mV dec–1) and the Ni foam (20.8 mV dec–1), indicating a rapid electron transfer rate and superior reaction kinetics (Fig. 3b and Supplementary Fig. 13). Regarding the UOR potential, the asymmetric Ni–O–Ti sites outperformed most of the previously reported Ni-based electrocatalysts (Supplementary Fig. 14 and Table 3).Fig. 3: Electrochemical performances of the atomically isolated asymmetric Ni–O–Ti sites.a LSV curves with 85% iR correction and (b) Tafel plots of the atomically isolated asymmetric Ni–O–Ti sites and connected symmetric Ni–O–Ni sites on Ti foam in 1.0 M KOH containing 0.33 M urea with a stirring rate of 500 rpm and 5 mV s–1. Freestanding Ti foam and Ni foam were used for comparison. The solution resistance is 2.1 ± 0.1 Ω. c The UOR selectivity of asymmetric Ni–O–Ti sites, symmetric Ni–O–Ni sites, and Ni foam as a function of the applied potential. Error bars represent the standard deviations derived from three distinct samples. d On-line MS for asymmetric Ni–O–Ti sites in 1.0 M KOH with 0.1 M CO(14NH2)2 and 0.1 M CO(15NH2)2. e The current density changes of asymmetric Ni–O–Ti sites after the introduction of 0.1 M NCO– and 0.1 M NO2– in 1.0 M KOH containing 0.33 M urea electrolyte. f Chronoamperometry measurements with the initial current density of 100 mA cm–2 in 1.0 M KOH containing 1.0 M urea, and the solution resistance is 2.0 ± 0.1 Ω.Regardless of their comparable UOR potentials in catalyzing the UOR, asymmetric Ni–O–Ti sites and symmetric Ni–O–Ni sites exhibited significant differences in their selectivity for N2 evolution. Systematic analyses using ion chromatography and on-line mass spectrometry within a potential range of 1.40 to 1.70 VRHE (Supplementary Figs. 15–18) revealed that the asymmetric Ni–O–Ti sites primarily generated innocuous N2 (1), achieving a N2 selectivity exceeding 99% (Fig. 3c, Supplementary Figs. 19–21 and Supplementary Movie 1)17,18,31,32. Concurrently, a notable H2 evolution rate of 22.0 mL h–1 was delivered at the coupled Pt cathode under a current density of 213 mA cm–2 at 1.40 VRHE (2–3). Symmetric Ni–O–Ni sites, however, exhibited a poor N2 selectivity of 61% under 1.40 VRHE, which decreased progressively to 56%, 14%, and 9% as the potential increased to 1.50, 1.60, and 1.70 VRHE, respectively (Supplementary Figs. 22, 23 and Table 4), accompanied by a substantial production of NCO– and overoxidized NOx– (4). Moreover, conventional Ni foam displayed a disappointingly low N2 selectivity ranging from 10% to 30%, illustrating that Ni–O–Ni sites disfavored N2 production (Supplementary Figs. 24, 25). Furthermore, the indispensable contribution of Ti within the asymmetric Ni–O–Ti sites to the high N2 selectivity of the UOR was confirmed by the poor N2 selectivity (below 10%) of single Ni atoms dispersed on N-doped graphene (Ni1@NC) (Supplementary Fig. 26).$$\begin{array}{cc}{{{{{\rm{Anode\; UOR}}}}}} \quad \quad & {{{{{\rm{CO}}}}}}{({{{{{\rm{NH}}}}}}_{2})}_{2}+6{{{{{{\rm{OH}}}}}}}^{-}\to {{{{{{\rm{N}}}}}}}_{2}+{{{{{{\rm{CO}}}}}}}_{2}+5{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}+6{{{{{{\rm{e}}}}}}}^{-}\end{array}$$
(1)
$$\begin{array}{cc}{{{{{\rm{Cathode\; HER}}}}}} \quad \quad & 6{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}+{6{{{{{\rm{e}}}}}}}^{-}\to 3{{{{{{\rm{H}}}}}}}_{2}+6{{{{{{\rm{OH}}}}}}}^{-}\end{array}$$
(2)
$$\begin{array}{cc}{{{{{\rm{Overall}}}}}} \quad \quad & {{{{{{\rm{CO}}}}}}({{{{{\rm{NH}}}}}}_{2})}_{2}+{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}\to {{{{{{\rm{N}}}}}}}_{2}+{3{{{{{\rm{H}}}}}}}_{2}+{{{{{{\rm{CO}}}}}}}_{2}\end{array}$$
(3)
$$\begin{array}{cc}{{{{{\rm{Overoxidation}}}}}} \quad \quad & {{{{{{\rm{CO}}}}}}({{{{{\rm{NH}}}}}}_{2})}_{2}+8{{{{{{\rm{OH}}}}}}}^{-}\to {{{{{{\rm{NCO}}}}}}}^{-}+{{{{{{{\rm{NO}}}}}}}_{2}}^{-}+6{{{{{{\rm{H}}}}}}}_{2}{{{{{\rm{O}}}}}}+{6{{{{{\rm{e}}}}}}}^{-}\end{array}$$
(4)
We subsequently utilized online mass spectrometry to analyze the mass-to-charge ratio of the produced N2 during UOR by feeding CO(14NH2)2 and CO(15NH2)2 in a 1:1 ratio. The observed mass spectral distribution of 14N2:15N2:14N15N at a 1:1:0 ratio offered compelling evidence of a dominant intramolecular N–N bond coupling reaction mediated by the asymmetric Ni–O–Ti sites (Fig. 3d)24. Control experiments indicated that the introduction of external 0.1 M NCO– and 0.1 M NO2– in the electrolyte led to the poisoning of asymmetric Ni–O–Ti sites, as evidenced by decreased current densities (Fig. 3e). To this end, the prevention of secondary NCO– and NOx– generation over the asymmetric Ni–O–Ti sites ensured the stability of the electrochemical system over 10,000 cycles and an extended period of 10 days with a well-maintained N2 selectivity of 99% during the laboratory test (Fig. 3f and Supplementary Figs. 27–30). A minor reduction in the current density was attributed to the depletion of urea as the UOR proceeded, which could be easily solved by refreshing the electrolyte. In contrast, the Ni–O–Ni sites and Ni foam exhibited more pronounced decreases in current density, retaining only 64.1% and 16.4% of their initial values, respectively.Reaction mechanism of the UORIn situ, Raman spectroscopy was utilized to clarify the role of atomically isolated asymmetric Ni–O–Ti sites in the selective oxidation of urea to N2 (Supplementary Fig. 31)33. The UOR on Ni-based electrocatalysts is known to follow a sequential electrochemical-chemical mechanism. That is, Ni first undergoes oxidation into high-valent Ni3+–O species. Subsequently, Ni3+–O reacts with the adsorbed urea to produce CO2, N2, and H2O, simultaneously regenerating the catalysts. This process was confirmed by the appearance of Ni3+–O on the asymmetric Ni–O–Ti sites at potentials ≥ 1.30 VRHE (Supplementary Fig. 32)34,35. A slightly greater potential was required for the formation of Ni3+–O species on the symmetric Ni–O–Ni sites and on the Ni foam (1.40 VRHE), which is consistent with the LSV curves of these materials. Upon the introduction of urea with a prototypical C⎓N stretching band at 1004 cm–1 (Fig. 4a)12, the Ni3+–O Raman peak rapidly diminished and completely vanished by 1.50 VRHE along with the emergence of adsorbed CO2 species (CO32–) at 1065 cm–1, highlighting Ni3+–O as the key species for the UOR (Supplementary Fig. 33)36. Interestingly, neither NCO– nor NOx– was observed on the asymmetric Ni–O–Ti sites (Supplementary Fig. 34), indicating that the C⎓N bond was well stabilized prior to intramolecular N–N coupling. In contrast, both the Ni–O–Ni sites and the Ni foam exhibited pronounced accumulations of NCO– (1221 and 1313 cm–1) and NO2– (1331 cm–1) species during the UOR, reflecting their poor selectivity for N2 evolution (Fig. 4b and Supplementary Fig. 35).Fig. 4: Mechanism of the UOR on the atomically isolated asymmetric Ni–O–Ti sites.In situ Raman spectra of (a) atomically isolated asymmetric Ni–O–Ti sites and (b) connected symmetric Ni–O–Ni sites at different potentials during the UOR operations. The distinct Raman peaks at 476 and 556 cm–1 are attributed to Eg bending vibration and the A1g stretching modes of Ni3+–O, respectively. c The FTIR spectra of urea solution and adsorbed urea on atomically isolated asymmetric Ni–O–Ti sites and connected symmetric Ni–O–Ni sites. In situ FTIR spectra during the UOR on (d) atomically isolated asymmetric Ni–O–Ti sites and (e) connected symmetric Ni–O–Ni sites at different potentials.To clarify the specific reaction pathways governed by the asymmetric Ni–O–Ti sites for the UOR, an in situ FTIR spectroscopy study was conducted37. Urea features ν(C = O) at 1670 cm–1, δ(NH2) at 1630 and 1600 cm–1, and νa(C⎓N) at 1466 cm–1 (Fig. 4c and Supplementary Fig. 36)38. As mentioned above, the C⎓N bond in urea possesses 30% hybridized double bonding characteristics due to the resonance between the NH2 and C = O groups. Upon interaction with symmetric Ni–O–Ni sites via NH2 groups, the C⎓N bond was weakened due to increased electron demand from donor N atoms, disrupting the resonance between the NH2 and C = O groups. This interaction is demonstrated by a redshift of νa(C⎓N) to 1445 cm–1 and a blueshift of ν(C = O) to 1694 cm–1, indicative of the weakening of the C⎓N bond and the strengthening of the C = O bond, respectively. These spectral shifts, in conjunction with the DFT simulations, corroborate the dual NH2 adsorption configuration of urea at symmetric Ni–O–Ni sites (Supplementary Fig. 37). In contrast, a blueshift of νa(C⎓N) to 1476 cm–1 was observed for urea adsorbed on the Ni–O–Ti sites, suggesting a strengthening of the C⎓N bond (Supplementary Fig. 38). Benefiting from the high oxygenophilicity of Ti, urea might bind to Ni–O–Ti sites in a new adsorption configuration via C = O→Ti interactions. With an increase in the potential to 1.30 VRHE, the IR peak assigned to urea gradually diminished, while a distinct peak attributed to the stretching vibration of the C–O single bond emerged at 1250 cm–1 (Fig. 4d)9,39. We believe that this peak is derived from the adsorption of the C = O group on Ni3+–O in the Ni–O–C = O→Ti configuration40. The peak at ~ 1380 cm–1 arose from CO32– in the electrolyte and gradually accumulated with increasing potential along with the physically adsorbed CO2 (2343 cm–1)41,42,43. The absence of NCO– or NOx– vibrational peaks in the spectra of asymmetric Ni–O–Ti sites across all potentials further confirms their high N2 selectivity. In contrast, symmetric Ni–O–Ni sites displayed an increasing NCO– peak at 2168 cm–1, associated with robust C⎓N bond cleavage of urea (Fig. 4e)44.Theoretical investigation of the UORBuilding on the experimental results and discussions, comprehensive density functional theory (DFT) calculations were conducted to clarify the UOR pathways governed by asymmetric Ni–O–Ti sites (Supplementary Fig. 39 and Supplementary Data 1). As previously demonstrated, the Ni–O–Ti sites favored interactions with the C = O group of urea with higher adsorption energy (– 0.71 eV) than did interactions with the NH2 groups (– 0.22 eV). The electrochemically generated Ni3+–O then bound to the C moiety of the C = O group (Supplementary Fig. 40), while the oxygenophilic Ti atom was associated with the O moiety, forming an asymmetric bidentate binuclear intermediate (Fig. 5a). Significantly, this urea adsorption configuration on Ni–O–Ti sites manifested an electron accumulation on the two C⎓N bonds (+ 0.02 e and + 0.03 e), indicating a strengthened resonance between the NH2 and C = O groups (Fig. 5b). This was in stark contrast to the Ni–O–Ni sites where urea adsorption through its NH2 groups with a larger adsorption energy of – 1.50 eV led to a decrease in the electron density on the C⎓N bonds (– 0.40 e and – 0.49 e, Fig. 5c). The strengthened resonance within the urea molecule further shortened the C⎓N bond length from 1.38 Å to 1.35 Å and increased the bond strength from 4.94 eV to 5.40 eV, according to crystal orbital Hamilton population (COHP) analysis (Fig. 5d)45,46,47,48. In contrast, the resonant C⎓N bond was elongated to a single C–N bond of 1.47 Å with a destabilized bond strength of 4.22 eV49. Thus, the asymmetric Ni–O–Ti sites might prevent premature breakage of the primary C⎓N bond prior to intramolecular N–N coupling.Fig. 5: Theoretical investigation of the UOR on the atomically isolated asymmetric Ni–O–Ti sites.a The adsorption energies of urea on asymmetric Ni–O–Ti sites and symmetric Ni–O–Ni sites in different configurations. Charge density difference of urea adsorbed on (b) asymmetric Ni–O–Ti sites and (c) symmetric Ni–O–Ni sites. The yellow and blue iso-surfaces with an iso-value of 0.004 a.u. represent spatial charge accumulation and depletion, respectively. d The COHP analysis of C–N bond in the free and adsorbed urea on asymmetric Ni–O–Ti and symmetric Ni–O–Ni sites. e The UOR process diagram and Gibbs free energy change on asymmetric Ni–O–Ti and symmetric Ni–O–Ni sites.The elemental steps of the UOR were then explored through broader theoretical calculations (Fig. 5e and Supplementary Figs. 41–43). Intriguingly, the urea adsorbed on the asymmetric Ni–O–Ti sites spontaneously underwent intramolecular N–N coupling during the initial dehydrogenation step, forming an *OCNH–NH2 intermediate while releasing 0.41 eV of energy. This pathway was more thermodynamically favorable than the alternative route, which involves the breaking of the C⎓N bond to form *ON + *OCNH2 intermediates, a reaction confronted with a large potential-independent energy barrier of + 1.03 eV. Subsequent stepwise dehydrogenation and desorption steps led to the proactive production of N2 with a relatively minor energy barrier of + 0.13 eV, consistent with the high UOR selectivity toward N2 on the asymmetric Ni–O–Ti sites. In contrast, at the symmetric Ni–O–Ni sites, urea necessitated a higher energy input (+ 0.34 eV) for the first dehydrogenation step to generate *NHCONH2. In the subsequent dehydrogenation stage, the generated *NHCONH was more inclined to undergo C⎓N bond cleavage into *HNO + *HNCO (Supplementary Fig. 44), a route that is more thermodynamically favorable by 0.23 eV over the formation of the *HN–NHCO intermediate via intramolecular N–N coupling, thus accounting for the excessive and uncontrolled oxidation of urea into NO2– and NCO– on conventional Ni-based catalysts during the UOR.Solar-powered UOR and H2 evolutionTo showcase the practical application of these asymmetric Ni–O–Ti sites, we designed a prototype UOR–HER device powered by a commercial monocrystalline Si solar panel (3.0 V, 49 cm2)50. The projected operation current of the designed system was estimated to be 183 mA according to the intersection of the J–V curve of the solar cell and the polarization curve of the electrolysis device (Fig. 6a). Compared with the OER, the potential required to reach a current density of 100 mA cm–2 on asymmetric Ni–O–Ti sites for the UOR was reduced by 0.28 V, resulting in an energy savings of ~ 17% for H2 production (Fig. 6b, Supplementary Fig. 45 and Supplementary Notes 2). Solar-powered UOR–HER electrolysis was then conducted on a clear summer day in Shanghai, China, using actual urine collected from public facilities and commercial Pt foil as the cathode electrode, delivering H2 and N2 production rates of approximately 3:1, with 75.7 and 25.2 mL h–1, respectively, and a solar-to-H2 conversion efficiency of 9.6% (Fig. 6c–e and Supplementary Fig. 46). The Faradaic efficiency of urea-to-N2 conversion and H2 production in this solar-powered reactor exceeded 99%. After 6 h of the electrochemical reaction, 96% of the N total in the urine was removed (Fig. 6f). In addition, the concentrations of NO3−, NO2−, and NH4+ were maintained at low levels throughout the other 4 cycles, revealing the high efficiency and stability of asymmetric Ni–O–Ti sites for practical UOR applications.Fig. 6: Solar-powered reactor design and performance.a J–V curve of the solar panel and the UOR polarization curves with the Ti foam containing atomically isolated asymmetric Ni–O–Ti sites as the anode and Pt foil as the cathode with a scan rate of 5 mV s–1. b LSV curves with 85% iR correction of the asymmetric Ni–O–Ti sites in 1.0 M KOH + 0.33 M urea and 1.0 M KOH for driving the UOR and OER, respectively. The on-line MS of H2 and N2/O2 collected from cathode and anode chambers, respectively, (c) in 1.0 M KOH + actual urine (urea: ~ 7 g L–1). d The digital image of the solar-powered reactor. e The collected H2 and N2 from actual urine. f Concentrations of N-total, NO3−, NO2− and NH3 during photo-electrochemical urine oxidation process.