Preliminary reactivity studies. TEMPO and related aminoxyl radicals have been widely used for electrochemical alcohol oxidation (EAO)21–23,25,26,34,35. In this study, we selected five aminoxyl radicals as model electrocatalysts (Fig. 2a) due to their varying oxoammonium/aminoxyl redox potentials resulting from differences in their R-functional groups. Nickel oxyhydroxide materials (NiOOH) are also promising heterogeneous electrocatalysts for EAO8,9,13,30,31,36–38. The resting state of NiOOH is Ni(OH)2 during the open-circuit conditions. We wondered whether the two electrocatalysts could be combined to achieve synergistic benefits for EAO. Benzyl alcohol (BA) was chosen as a model substrate for EAO, and the BA oxidation reactions were conducted under constant-potential conditions (1.45 V vs. RHE) in a continuous-flow parallel-plate electrolyzer with electrode area of 10 cm2, in which Ni-based materials as the anode and nickel foam as the cathode (see section Methods for details). Gas chromatography (GC) was utilized to monitor and quantify the concentration changes of the organic compounds. The synthesis and characterization of the NiM-LDH materials are elaborated below and in the Supporting Information (see section Methods for details).
Attempts to perform direct oxidation of BA at 1.45 V vs. RHE with a Ni(OH)2-impregnated graphite-felt electrode resulted in only low yield of the target product benzaldehyde (PhCHO) (5%, Table 1, entry 1). Better results were obtained with 5 mol% AcNH-TEMPO (ACT) rather than with a heterogeneous electrocatalyst, with the yield of PhCHO increasing to 38% (entry 2). The yield increased further when both Ni(OH)₂ and ACT were used together (52%, entry 3), raising the prospect of a synergistic effect that exceeds the sum of the two individual catalysts. Analysis of heterogeneous Ni(OH)2-based materials that feature other metals within the lattice were only marginally better as heterogeneous electrocatalysts (e.g., entry 4); however, a NiV-LDH material with a Ni0.67V0.33 stoichiometry (described further below) paired with ACT led to nearly quantitative yield of PhCHO (entry 5). Other NiM-LDH materials with ACT were not as effective as the NiV-LDH/ACT catalyst system (M = Mn, Fe, Co, Cu; entries 6–9). In addition, ACT is superior to TEMPO (entry 10), 4-HO-TEMPO (entry 11), 4-MeO-TEMPO (entry 12) and 4-oxo-TEMPO (entry 13).
Computational analysis of aminoxyl interactions with NiM-LDH materials. Density functional theory (DFT) calculations were performed to investigate the electrocatalytic reactivity. Figure 2a displays the reactivity of various aminoxyl radicals on a clean NiOOH electrode surface. Adsorption energies for H-TEMPOH, 4-HO-TEMPOH, 4-MeO-TEMPOH, AcNH-TEMPOH (ACTH), and 4-oxo-TEMPOH were calculated to be -0.44 eV, -0.96 eV, -0.15 eV, -0.73 eV, and − 0.68 eV, respectively (Supplementary Fig. 2). The adsorption energy for the above components follows the order of 4-HO-TEMPOH > AcNH-TEMPOH > 4-oxo-TEMPOH > H-TEMPOH > 4-MeO-TEMPOH. Despite the strongest interaction between 4-HO-TEMPOH and NiOOH surface, the oxidative activity of 4-HO-TEMPO is attenuated in alkaline environments, attributed to deprotonation of the HO- group, reducing in the electron density of the aminoxyl radical and impairing its oxidative capability. Therefore, the ACT shows exhibits good adsorption and superior electrocatalytic activity. Further calculations were performed to assess ACTH adsorption to various NiM-LDH materials, with M = V, Mn, Fe, Co and Cu. The V-doped NiOOH material exhibited the strongest adsorption of ACTH (Fig. 2b), with an adsorption energy calculated to be -2.04 eV, and the NiV-LDH material binds ACT most strongly among the different aminoxyls (Fig. 2c). The amide carbonyl group of ACT binds to an exposed Ni ion on the NiV-LDH surface, with a calculated Ni-O bond length of 1.98 Å.
Synthesis and characterization of NiM-LDH materials. The heterogeneous NiM-LDH electrocatalysts (M = V, Mn, Fe, Co, Cu) were distributed uniformly on graphite felt (GF) via a hydrothermal method39, as schematically illustrated in Fig. 3a. NiV-LDH/GF electrocatalysts with different Ni/V molar ratios were obtained by controlling the dosage of nickel nitrate and vanadium (III) chloride during the hydrothermal synthesis process. Ni(OH)2/GF was also synthesized by a similar procedure but without V dopants. The scanning electron microscopy (SEM) displayed that the Ni0.67V0.33-LDH has a three-dimensional (3D) nanosphere morphology composed of ultrathin double hydroxide nanosheets with a diameter of approximately 600 nm (Fig. 3b, c), sharply contrasting the smooth surface of the pristine GF (Supplementary Fig. 3). The Ni(OH)2 also shows that the entire surface of the GF is uniformly coated with the dense nanosheets (Supplementary Fig. 4). Such porous structure exposes large number of active sites, facilitating penetration and diffusion of electrolytes and increasing the effective contact between the electrode and electrolytes during the EAO process40,41. Similarly, other Ni-based LDHs were evenly distributed on the surface of the GF (Supplementary Fig. 5). Transmission electron microscopy (TEM) images illustrate that the Ni0.67V0.33-LDH has a nanosphere structure consisting of thin interconnected nanosheets (Fig. 3d, e) that are also reflected in the SEM results. The lattice fringe spacing of the Ni0.67V0.33-LDH is approximately 0.205 nm42–44, as revealed by the high-resolution TEM (HRTEM) in Fig. 3f. The corresponding selected area electron diffraction (SAED) image in the inset of Fig. 3f shows that the Ni0.67V0.33-LDH contains multiple crystal planes in the form of circles of bright dots43,45. The energy dispersive X-ray spectrum (EDX) analysis reveals that the Ni0.67V0.33-LDH contains Ni, V, C and O as the main elemental components, with a V:Ni atomic ratio of approximately 1:2.38 (Supplementary Fig. 6). The elemental analysis in Fig. 3g also confirms that the elements Ni, V, and O are evenly distributed in the Ni0.67V0.33-LDH. The mass content of Ni and V was determined using inductively coupled plasma spectroscopy (ICP) to be 28.44% and 13.21%, respectively.
The powder X-ray diffraction (XRD) patterns of Ni0.67V0.33-LDH are shown in Fig. 3h. To avoid interference from the strong signal of the GF, powders synthesized by the same method were characterized. The diffraction peaks of Ni(OH)2 can be assigned to the standard brucite (space group: P3m1) crystal phase of β-Ni(OH)2 (JCPDS No. 73-1520)46,47, without any other peaks. After incorporating V into the structure of Ni(OH)2, the diffraction peaks of Ni0.67V0.33-LDH centered at 11.46°, 23.26°, 33.46°, 34.41°, 38.76° and 59.80°, corresponding to the (003), (006), (101), (012), (015) and (110) facets of hexagonal α-Ni(OH)2 (JCPDS No. 38–0715) with larger interlayer spacing17,48,49, respectively. This result indicates that the incorporation of V transforms Ni(OH)2 from β phase to α phase44. In addition, Brunauer-Emmett-Teller (BET) analysis probed the specific surface area of Ni0.67V0.33-LDH (Supplementary Fig. 7). The nitrogen adsorption-desorption curves of Ni0.67V0.33-LDH display a type IV isotherm with a hysteresis loop in the range of relative pressure (P/P0 = 0.45-1.0) and pore size in the range of 2 to 8 nm, indicating that Ni0.67V0.33-LDH is a mesoporous material. Moreover, the Ni0.67V0.33-LDH has a larger BET surface area (9.66 m2 g− 1) compared to β-Ni(OH)2 (2.81 m2 g− 1), demonstrating that V-doping alters the morphology of the nanospheres and is expected to improve the EAO efficiency by facilitating the mass transfer of the reactants and electrolytes. The surface chemical composition and electronic state of the elements in the Ni0.67V0.33-LDH and β-Ni(OH)2 powders were investigated by using X-ray photoelectron spectroscopy (XPS) (Supplementary Fig. 8a). The Ni 2p XPS spectrum of Ni0.67V0.33-LDH and β-Ni(OH)2 showed two Ni 2p3/2 peaks (Fig. 3i), including the dominant Ni2+ species and a Ni3+ species with a central binding energy of 855.9 and 858.1 eV, respectively48,50. Moreover, the V 2p3/2 signal exhibits two peaks at 516.6 and 517.5 eV (Fig. 3j), indicating the existence of a large number of V4+ species and a small number of V5+ species, respectively17. The latter data confirm that V is doped into the Ni0.67V0.33-LDH. The O 1s peaks observed for Ni0.67V0.33-LDH at 531.1 and 532.5 eV are attributed to metal-oxygen (Ni-O-V) bonds and adsorbed H2O, respectively (Supplementary Fig. 8b).
Analysis of synergistic NiM-LDH/ACT electrocatalytic alcohol oxidation. To evaluate the electrochemical performance of NiM-LDH/ACT for EAO, linear sweep voltammetry (LSV) analysis was carried out in a batch reactor using a typical three-electrode system (0.5 M Na2CO3 aqueous electrolyte with pH 11.5). As displayed in Fig. 4a, the LSV of Ni0.67V0.33-LDH electrocatalyst exhibits significantly higher electrocatalytic activity compared to other NiM-LDHs (M = Mn, Fe, Co, Cu). The oxidation peak at 1.53 V vs. RHE was identified as the Ni2+/Ni3+ oxidation peak related to the oxidation of Ni(OH)2 to NiOOH51,52, indicates that the incorporation of V into Ni-based LDH can effectively promote the oxidation of Ni2+ sites. Figure 4b demonstrates that the V content in the NiV-LDHs electrocatalysts has a volcano-like relationship with the oxidation current density, with a Ni/V atomic ratio of ~ 2:1 (Ni0.67V0.33-LDH) exhibiting the highest current density (Supplementary Fig. 9).
Use of 5 mol% ACT reveals an oxidation peak at 1.62 V vs. RHE (Fig. 4c), corresponding to oxidation of ACT to the oxoammonium species, ACT+. The oxidation peak intensity is significantly enhanced when both Ni0.67V0.33-LDH and ACT are present, reaching up to 130 mA/cm2 at 1.58 V vs. RHE. Flow oxidation studies conducted with benzyl alcohol (BA) reveal complementary results. Use of 100 mM BA reveals a much higher current density (336 mA cm− 2) for the cooperative Ni0.67V0.33-LDH/ACT electrolysis system relative to ACT alone (153 mA cm− 2) or Ni0.67V0.33-LDH alone (3 mA cm− 2) at 1.45 V vs. RHE in the flow electrolyzer (Fig. 4d, e). The corresponding Tafel slope of Ni0.67V0.33-LDH/ACT is 47 mV dec–1 (Fig. 4f), which is significantly lower than that of the ACT (64 mV dec− 1) or Ni0.67V0.33-LDH (88 mV dec–1), reflecting more favorable reaction kinetics for BA oxidation, relative to the two individual catalysts, Ni0.67V0.33-LDH and ACT. Electrochemically active surface area (ECSA) measurements were performed by systematically measuring CVs within the non-faradaic potential range to quantify the double-layer capacitance (Cdl, Supplementary Fig. 10). The NiV-LDH shows an increase in bilayer capacitance relative to Ni(OH)2, 90.8 mF cm− 2 vs. 31.2 mF cm− 2 (Supplementary Fig. 11), suggesting that increase access to surface sites in the Ni0.67V0.33-LDH material at least partially contributes to its improved electrocatalytic performance.
A series of chronoamperometry experiments were conducted at different potentials (1.30–1.50 V vs. RHE in a 0.5 M Na2CO3 electrolyte) in the continuous-flow electrolyzer to investigate further the origin of Ni0.67V0.33-LDH/ACT cooperativity. Figure 4g shows that the target product, benzaldehyde (PhCHO), was produced with a selectivity of 99% at a voltage of 1.45 V vs. RHE using the cooperative Ni0.67V0.33-LDH/ACT electrode. Nearly 1930 C (2 F/mol) of charge is required to completely convert BA into PhCHO (Supplementary Fig. 12). Benzoic acid (PhCOOH) formation is negligible during the process (Supplementary Fig. 13). Independent testing of Ni0.67V0.33-LDH and ACT for EAO performance at 1.45 V vs. RHE leads to only 9% and 38% yield of PhCHO, respectively (Fig. 4h), confirming the synergistic benefits of Ni0.67V0.33-LDH and ACT for alcohol oxidation. The selectivity and yield of PhCHO remains consistently above 95%, even after 10 consecutive electrochemical oxidation cycles (Fig. 4i). The morphology and structure of Ni0.67V0.33-LDH were not significantly altered after electrochemical oxidation experiments, as shown by SEM, TEM, and elemental mapping (Supplementary Figs. 14 and 15). EDX analysis after the reaction revealed a slight loss of Ni and V and an increase in O content compared to the pre-reaction Ni0.67V0.33-LDH electrocatalyst (Supplementary Fig. 16); however, high-resolution XPS spectra of Ni 2p and V 2p exhibted a slight increase in Ni3+ and V5+ after EAO compared to the as-prepared Ni0.67V0.33-LDH material (Supplementary Fig. 17). These observations are rationalized by partial reconstruction of the LDH material during EAO.
Mechanistic analysis of NiV-LDH/ACT co-catalytic alcohol oxidation. A series of experiments and computations were performed to probe mechanistic features of the EAO reaction. In situ electrochemical impedance spectroscopy (EIS) was conducted at various potentials to investigate the interfacial behavior of Ni0.67V0.33-LDH/ACT8,13,36. For Ni0.67V0.33-LDH in 0.5 M Na2CO3, the results reveal a shift to lower phase and higher frequencies with increasing applied potentials, reflecting an increase in reaction rates (Supplementary Fig. 18). As shown in Fig. 5a, the phase angle decreases significantly and frequency shifts higher after the addition of ACT and BA (blue line), indicating a faster oxidation rate and faster kinetics of EAO of BA. Meanwhile, the corresponding Nyquist plots displayed approximately vertical lines in 0.5 M Na2CO3 (Supplementary Fig. 19), suggesting a high charge transfer resistance for the Ni0.67V0.33-LDH electrode at 1.37 V vs. RHE (Fig. 5b). The Nyquist plots show a smaller semicircle when both ACT and BA are present in the electrolytes (blue line), reflecting minor charge transfer resistance that is consistent with efficient Ni0.67V0.33-LDH/ACT-mediated EAO53. In addition, the Bode plot and Nyquist plot were also analyzed for the β-Ni(OH)2/ACT in the batch reactor (Supplementary Figs. 20 and 21), the results revealed that the larger semicircle of the Nyquist plot compared with Ni0.67V0.33-LDH/ACT, implied that the transfer resistance can be reduced after the introduction of V. In situ Raman experiments were performed at various potentials to monitor the changes in the surface structure of Ni0.67V0.33-LDH and β-Ni(OH)2 electrodes (Fig. 5c-f and Supplementary Fig. 22). With β-Ni(OH)2, no discernible change is observed upon increasing the potential up to 1.37 V vs. RHE (Fig. 5c), while two new peaks become evident at 473 and 557 cm− 1 upon increasing the potential to 1.47 V vs. RHE8,54. These peaks are assigned to surface Ni3+-O vibrations upon oxidation of Ni(OH)2 to NiOOH. Ni0.67V0.33-LDH exhibits a Raman signal at approximately 767 cm− 1 that is absent in the β-Ni(OH)2 catalyst and is attributed to a V-O stretch (Fig. 5d). Increasing the applied potential to 1.37 V vs. RHE leads to gradual enhancement of the 767 cm− 1 peak, in addition to emergence of the NiOOH peaks at 473 and 557 cm− 1, indicating that NiOOH forms at a lower potential in to form in Ni0.67V0.33-LDH catalysts at a lower potential than in β-Ni(OH)2, thus the introduction of V facilitates the formation of NiOOH, consistent with the LSV and EIS. The peaks of NiOOH were attenuated at 1.37 V vs. RHE after the addition of ACT compared to without ACT in situ Raman spectroscopy (Fig. 5e), indicating that Ni3+ was initially formed on the surface of Ni0.67V0.33-LDH, then quickly captured the electrons, and partially transformed back into Ni2+ again55. Because the reduction of the Ni3+ by ACT forms ACT+, which can oxidize BA to PhCHO, these results align with the synergistic benefits of NiV-LDH and ACT for EAO. DFT calculations illuminate the increase in electron density upon electron transfer from an adsorbed ACT to a Ni3+ site (Fig. 5g, h). Taken together, these results raise the possibility that enhanced formation of the more oxidizing and Lewis acidic Ni3+ by the V ions in the NiV-LDH promotes ACT or ACTH adsorption and oxidation to ACT+ species, and the adsorbed ACT+ species could then promote rapid alcohol oxidation (Fig. 5i).
Electrochemical oxidation of other alcohols with the synergistic NiV-LDH/ACT co-catalyst system. EAO performance was then evaluated with Ni0.67V0.33-LDH, ACT, and Ni0.67V0.33-LDH/ACT with a series of different alcohol substrates using a recirculating-flow parallel-plate electrolyzer (Fig. 6). Due to the different solubilities of the substrates, we chose different ratios of aqueous sodium carbonate and acetonitrile to ensure that the substrates completely dissolved and the electrolyte retained good conductivity. The cooperative Ni0.67V0.33-LDH/ACT electrocatalyst shows superior performance relative to Ni0.67V0.33-LDH and ACT used independently in the oxidation of primary alcohols (1a-7a) to the corresponding aldehydes, affording yields of ≥ 87% for 1b-7b in ≤ 46 min. Oxidation of the pharmaceutically relevant steroid 19-hydroxyandrost-4-ene-3,17-dione, 8a, with the Ni0.67V0.33-LDH/ACT co-catalyst generated the aldehyde 8b in 96% yield, which is significantly better than Ni0.67V0.33-LDH (2%) and ACT alone (30%). Efforts then focused on synthesis of a precursor of tibolone [(7α,17α)-17-hydroxy-7-methyl-19-norpregn-5(10)-en-20-yn-3-one]. The substrate 9a (7α-methyl-19-aldehyde-4-androstene-3,17-dione) was evaluated on 50 mmol scale. Despite the large molecular weight and poor solubility of the substrate, a current density of 150 mA/cm2 (based on the electrode two-dimensional surface area) was accessible and showed stable anode potential of 1.4–1.6 V vs. RHE during 15 A constant current electrolysis with a flow rate of 2000 mL/min (Fig. 6B-a). Time-course analysis by HPLC showed that the concentration of 9a decreases steadily with parallel formation and conversion of intermediates 9a-1 and 9a-2, ultimately converging on the oxidation product 9b (Fig. 6B-b). High selectivity and yield of 9b (95% and 94%, respectively) were achieve within 25 min.
We then focused in increate the reaction productivity by increasing the reactor volume and establishing a large-scale electrochemical synthesis system for oxidation of sterol 8a to 8b with H2 production at the counter electrode. An anode consisting of Ni0.67V0.33-LDH/GF and a Ni foam cathode were integrated within three electrolyzers, with two-dimensional electrode surface areas of 10, 100, and 400 cm2 (Fig. 7a, b). Computational fluid dynamics (CFD) simulation revealed that enlarging the electrolyzer caused non-uniform flow rates within the reactor, particularly at the corners of the flow channels where the fluid flow is essentially absent (Fig. 7c). Introducing a cavity structure in the filled electrodes enables significantly improved fluid flow, resulting in a more uniform flow rate distribution (Fig. 7d). The LSV curve in Fig. 7e shows that the onset potential for the oxidation of sterol 8a (200 g) is 1.2 V vs. RHE, significantly lower than that of OER (Fig. 7f). Optimization of the electrolysis method, including the reaction current, electrolyte flow rate and electrolyte temperature, led to high productivity in formation of the steroid carbonyl product 8b from 8a (Fig. 7g). The resulting product exhibited a selectivity of 94% during a constant current electrolysis of 60 A in the 400 cm2 electrolysis cell (Supplementary Fig. 23). The productivity associated with this reactor is 243 g/h with an energy consumption of approximately 0.71 kWh/kg. This reaction was repeated on hectogram-scale (182.7 g), affording 8b in 91% isolated yield on the basis of HPLC and 1H NMR analysis (Supplementary Figs. 24–27). Collectively, these results demonstrated the utility of this cooperative Ni0.67V0.33-LDH/ACT-mediated for larger scale EAO applications.