Electrocatalytic Reduction of Dinitrogen to Ammonia with Water as Proton and Electron Donor Catalyzed by a Combination of a Tri-ironoxotungstate and an Alkali Metal Cation

The electrification of ammonia synthesis is a key target for its decentralization and lowering impact on atmospheric CO2 concentrations. The lithium metal electrochemical reduction of nitrogen to ammonia using alcohols as proton/electron donors is an important advance, but requires rather negative potentials, and anhydrous conditions. Organometallic electrocatalysts using redox mediators have also been reported. Water as a proton and electron donor has not been demonstrated in these reactions. Here a N2 to NH3 electrocatalytic reduction using an inorganic molecular catalyst, a tri-iron substituted polyoxotungstate, {SiFe3W9}, is presented. The catalyst requires the presence of Li+ or Na+ cations as promoters through their binding to {SiFe3W9}. Experimental NMR, CV and UV–vis measurements, and MD simulations and DFT calculations show that the alkali metal cation enables the decrease of the redox potential of {SiFe3W9} allowing the activation of N2. Controlled potential electrolysis with highly purified 14N2 and 15N2 ruled out formation of NH3 from contaminants. Importantly, using Na+ cations and polyethylene glycol as solvent, the anodic oxidation of water can be used as a proton and electron donor for the formation of NH3. In an undivided cell electrolyzer under 1 bar N2, rates of NH3 formation of 1.15 nmol sec–1 cm–2, faradaic efficiencies of ∼25%, 5.1 equiv of NH3 per equivalent of {SiFe3W9} in 10 h, and a TOF of 64 s–1 were obtained. The future development of suitable high surface area cathodes and well solubilized N2 and the use of H2O as the reducing agent are important keys to the future deployment of an electrocatalytic ammonia synthesis.


Figure S2. Cyclic Voltammetry of Aldehyde
The measurement conditions were 6 mL Dry THF containing with 0.1M TBAPF6 and 110 mM aldehyde.The solution was purged for 30 min with N2.A glassy carbon disc working electrode, a platinum wire counter electrode, and a Fc/Fc + reference electrode were used at a scan rate of 100mV/s Figure S3.UV-vis spectra of TBA{SiFe3W9}, Li + and N2 before and after electrolysis There are some changes in the intensities of the peaks that in comparison with Figure 4d indicates the presence of a residual amount of 1-electron reduced species.Computed molecular electrostatic potential mapped onto an isodensity surface of 0.0004 for {SiFe3W9O37}.Lithium-center coordination modes are also included in bridge (green) and terminal (blue) oxygens.Three lithium cations preferentially approach to the three nucleophilic wells near [Fe3O3], but only one is localized in a well generated by [W4O4].We noticed that this Li + cation is also coordinated to a ClO4 -, which is in turn attached to other Li + in terminal oxygens.(c) Radial distribution functions (RDFs, g(r)) (in black line) between the {SiFe3W9O37} (Si or Fe as reference) and, from left to right, N of TBA, Li + , Cl of ClO4 -and O of THF.Integration of the g(r) (in red line) is also included.   + 10Li + + 3ClO4 -+ continuum dielectric.Orbital energies are in eV.Fe, W and O labels represent the atoms with higher contribution to the molecular orbitals.The reduction potential of a polyoxometalate depends on the absolute energies of the LUMOs.In the gas phase, the molecular orbitals of a polyoxometalate are, in general, very high in energy because of the negative charge of the anion.In solution, the solute polyoxometalate orbitals are much lower in energy due to the electric field created by solvent molecules and counter cations.For highly charged compounds such as the {SiFe III 3W9O37 7-} anion under consideration here, the continuum solvent methods were unable to correctly simulate the environment (solvent + counterions) effects.The consequence is that the frontier molecular orbital energies are excessively high.In addition, and importantly, under the present experimental conditions, MD simulations show that several Li + ions are in direct contact with the polyoxometalate, introducing an extra stabilization of the polyoxometalate, which cannot be reproduced by an implicit solvation method.Therefore, addition of a Li + salt to the THF solution drastically changes the properties of the polyoxometalate anion, the redox activity being one of the most affected properties.In the absence of THF ligands, all attempts to bind a N2 to one of the Fe(II) centers failed as it was impossible to find an energy minimum in the region close to 1.9 Å.We find a decrease in energy only when N2 moves away from the metallic center.Reaction conditions: 10 mL PEG-400 containing 0.1 M TBAPF6, with or without 0.5 mM TBA{SiFe3W9}, 25 mM NaClO4 with 1 vol% water under 1 bar N2 for 3 h using a copper wire working electrode, a platinum wire counter electrode, and a Ag/AgCl reference electrode.For the 15 N2 experiment, it should be noted that due the high viscosity of PEG-400, and its low volatility, excellent results in degassing the solvent to remove 14 N2 were obtained by purging with He for 30 min at 60 °C, followed by the introduction of 15 N2.The residual 14 NH3 peak is attributed to the isotopic purity (98%) of the 15 N2 used and possibly other small contaminations.The coupling constant for 14 NH3 is 53 Hz; The coupling constant for 15 NH3 is 72 Hz.The reaction was carried out in and undivided cell electrolyzer, consisting of a 0.13 cm 2 Ni mesh cathode, a stainless-steel anode loaded with 2 mL PEG-400 containing 0.5 mM Na{SiFe3W9}, 1 vol% H2O, and 0.1 M NaCF3SO3 under 1 bar N2 operated at -1.3 V versus SHE.The reaction was carried out in and undivided cell electrolyzer, consisting of a 0.13 cm 2 Cu foam cathode, a stainless-steel anode loaded with 2 mL PEG-400 containing 0.5 mM Na{SiFe3W9}, 1 vol% H2O, and 0.1 M NaCF3SO3 under 1 bar N2 operated at -1.3 V versus SHE for 10 h.In an undivided cell electrolyzer, consisting of a 0.25 cm 2 Cu foil cathode, a stainless-steel anode loaded with 2 mL PEG-400 containing 0.5 mM Na{SiFe3W9}, 1 vol% H2O, and 0.1 M NaCF3SO3 under 1 bar N2 operated at -1.3 V versus SHE, the electrolyzer that yielded ~900 nmol NH3.The current obtained is shown in black.After removal of the cathode after 2 h and a gentle wash was the reaction was continued with the same cathode for another 2 h, red line.

Figure S4 .
Figure S4.Calibration of Various Electrodes in dry THF in the Presence of 0.1 M TBAPF6 as Electrolyte using Pt wires as Counter and Reference Electrodes.Fc/Fc + was measured using a 4 mM solution of Ferrocene with a Pt disk working electrode.Ag wire an Ag/AgCl were measured using them as working electrodes.

Figure S5 .
Figure S5.(a) Electrostatic isopotential surface showing the most negative potential (nucleophilic) wells of the {SiFe3W9O37} anion.(b) Computed molecular electrostatic potential mapped onto an isodensity surface of 0.0004 for {SiFe3W9O37}.Lithium-center coordination modes are also included in bridge (green) and terminal (blue) oxygens.Three lithium cations preferentially approach to the three nucleophilic wells near [Fe3O3], but only one is localized in a well generated by [W4O4].We noticed that this Li + cation is also coordinated to a ClO4 -, which is in turn attached to other Li + in terminal oxygens.(c) Radial distribution functions (RDFs, g(r)) (in black line) between the {SiFe3W9O37} (Si or Fe as reference) and, from left to right, N of TBA, Li + , Cl of ClO4 -and O of THF.Integration of the g(r) (in red line) is also included.

Figure S10 .
Figure S10.Representation of the three MOs occupied upon the 3e-reduction in the THF-containing model (see main text for details).Two electrons are delocalized at the three Fe centers, with a lower contribution of the Fe bound more strongly to one of the THF ligands, and the third electron is delocalized among W centers.

Figure S11 .
Figure S11.Schematic molecular orbital diagram for the 3-electron-reduced systems: I) {SiFe II 3W9O37} 7-+ 10Li + + 3ClO4 -+ continuum dielectric and II) {(THF)3SiFe II 2Fe III W VI 8WV O37} 7-+ 10Li + + 3ClO4 -+ continuum dielectric.The binding of three solvent molecules to the polyoxometalate induces a significant change in its electronic structure.In particular, the three lowest occupied (beta) d(Fe) molecular orbitals are destabilized by the presence of THF ligands, causing the transfer of one electron to the polyoxotungstate framework.Orbital energies are in eV.Fe, W and O labels represent the atoms with higher contribution to the molecular orbital.

Figure S12 .
Figure S12.The curves show how the energy (blue) and distance (black) of Fe•••N2 change during the optimization process when trying to coordinate a N2 molecule to the 3-electron reduced catalyst.In the absence of THF ligands, all attempts to bind a N2 to one of the Fe(II) centers failed as it was impossible to find an energy minimum in the region close to 1.9 Å.We find a decrease in energy only when N2 moves away from the metallic center.

Figure
Figure S13.1 H NMR (selgpse, 500.08MHz) after 5 h CPE in an electrolyzer: 0.1 M TBAPF6, 0.5 mM {SiFe3W9}, 25 mM LiClO4 in THF with 1 vol% ethanol as proton donor under 1 bar 14 N2 (blue) or 15 N2 (red) using a copper foil as working electrode, a stainless-steel counter electrode.The residual 14 N peaks in the 15 N2 experiment is associated with the isotopic purity of the 15 N2 used, experimental difficulties encountered in purging 14 N2 from volatile THF, (see FigureS16where PEG-400 was solvent and no purging difficulties were encountered) and possibly atmospheric contamination by 14 NH3.The coupling constant for 14 NH3 is 53 Hz; The coupling constant for 15 NH3 is 72 Hz.

Figure S14 .
Figure S14.Calibration of Various Electrodes in dry PEG-400 in the Presence of 0.1 M TBAPF6 as Electrolyte using Pt wires as Counter and Reference Electrodes.Fc/Fc + was measured using a 4 mM solution of Ferrocene with a Pt disk working electrode.Ag wire an Ag/AgCl were measured using them as working electrodes

Figure S17 .
Figure S17.Current versus time profile for N2 reduction on Cu.The reaction was carried out in and undivided cell electrolyzer, consisting of a 0.13 cm 2 Cu foam cathode, a stainless-steel anode loaded with 2 mL PEG-400 containing 0.5 mM Na{SiFe3W9}, 1 vol% H2O, and 0.1 M NaCF3SO3 under 1 bar N2 operated at -1.3 V versus SHE for 3 h.

Figure S18 .
Figure S18.Current versus time profile for N2 reduction on Ni.

Figure S19 .
Figure S19.Current versus time profile for N2 reduction on Cu for a 10 h reaction.

Figure S21 .
Figure S21.Recovered Cathode Experiment.In an undivided cell electrolyzer, consisting of a 0.25 cm 2 Cu foil cathode, a stainless-steel anode loaded with 2 mL PEG-400 containing 0.5 mM Na{SiFe3W9}, 1 vol% H2O, and 0.1 M NaCF3SO3 under 1 bar N2 operated at -1.3 V versus SHE, the electrolyzer that yielded ~900 nmol NH3.The current obtained is shown in black.After removal of the cathode after 2 h and a gentle wash was the reaction was continued with the same cathode for another 2 h, red line.

Figure S25 .
Figure S25.Thermogravimetric analysis plot of TBA{SiFe3W9}.The weight loss between 250 and 450 °C is attributed to the pyrolysis of TBA.Thus, leading to a 10:1 ratio of ratio of TBA:α-[SiW9O37{Fe(H2O)}3] and formulation of TBA{SiFe3W9} as TBA7[α-[SiFe III 3(H2O)3W9O37]•3TBA.Note that the excess of TBA has no bearing on the electrochemical results since this cation is present in excess as electrolyte.

Figure S26 .
Figure S26.Electrochemical setup and gas feed circulation.

Table S4 .
Faradaic efficiencies in the three-electrode setup at a function of time, Figure8a.