Green synthesis of ammonia from steam and air using solid oxide electrolysis cells composed of ruthenium‐modified perovskite catalyst

The green electrochemical synthesis of ammonia (NH3) through solid electrolysis has been intensively investigated. This research reported an improvement of NH3 production rate using in situ exsolution of ruthenium (Ru) atoms into lanthanum strontium chromium ferrite perovskite (LaxSr1−xCryFe1−yO3−δ) catalyst. The in situ Ru exsolution was achieved by reducing the optimized stoichiometric La0.33Sr0.67Cr0.33Fe0.52Ru0.15O3−δ (LSCrFRu) powders obtained in 10‐vol% H2/Ar at 800°C for 1.0 h. These Ru nanoparticles (NPs) were embedded on the surface of the LSCrFRu matrix, evenly distributed with a size varying from 4.4 ± 0.5 nm. With the exsolution of Ru atoms, greater oxygen vacancies were formed in ex‐LSCrFRu and gadolinium‐doped ceria (GDC) composite than those in LSCrF‐GDC electrocatalyst, beneficial to N2 gas adsorption and triple bond cleavage. These Ru‐modified LSCrFRu‐GDC catalysts showed a synthesis rate of 4.73 × 10−10 mol s−1 cm−2 at 550°C under 1.6 V, doubling the rate using LSCrF‐GDC catalyst. The improved ammonia synthesis kinetic is mainly attributed to embedded Ru NPs and the increased oxygen vacancies formed during the in situ exsolution process. More active sites and higher activity for H2O and N2 adsorption and activation collectively advanced facile transportation of O2− that further promotes the cleavage of covalent bonds in H2O, providing more H+ for the hydrogenation in the nitrogen reduction reaction. This research will open a new paradigm for the electrochemical synthesis of ammonia with mitigating the drawbacks of traditional NH3 production.


| INTRODUCTION
The synthesis of ammonia (NH 3 ) from nitrogen (N 2 ) and hydrogen (H 2 ) is considered the major industrial step of the last century. 1 The process was developed by Haber-Bosch and Mittasch in 1913 and is essentially used today with a few refinements. 2 The benefits of the Haber-Bosch process are that it enables the industrial production of ammonia 3 to be used as liquid fertilizer, 4 industrial refrigerant, 5 and as an intermediate in the synthesis of solid fertilizers (ammonium nitrate), 6 explosives, 7 and nitrogencontaining (hydrazine, hydroxylamine, and acrylonitrile) intermediates used in the chemical industries. 8 The role of fertilizer greatly accelerated grain (corn, wheat, and rice) production, which feeds approximately 80% of the world population, underscoring the importance of ammonia as a fertilizer and potential clean fuel. 9 The global ammonia production was 236 million tons in 2021, reaching 290 million tons by 2030. 10 The drawbacks of the Haber-Bosch process are the requirement of high pressure, thermal energy, and extraction of hydrogen from steam-methane reforming. As these are energy-intensive processes, coal is presently used to generate electricity and hydrogen, accounting for almost 1.5% of global CO 2 emissions or 1.5-1.6 tons CO 2-eq per ton of NH 3 . 11 The demand for ammonia is likely to increase due to the growing population and transition to sustainable energy resources or energy carriers. 12 Ammonia is attractive because of its high energy density as a fuel (22.5 MJ kg −1 at 1 bar and −33°C), generating nitrogen and water as a byproduct or in liquified form as a long-distance transport of hydrogen. 13 The conflict in Eastern Ukraine has also demonstrated that capital plants cannot easily be dismantled and transported out of a conflict zone, coupled with the European Union (EU) import ban on hydrocarbons from the Russian Federation means that liquefied ammonia from Qatar/ Australia could supplement the shortfall imposed by their EU import ban of coal, natural gas, and petroleum. 14 Electrochemical-driven ammonia solution minimizes many of the challenges of the Haber process in that the process is scalable due to synthesis at ambient pressure. [15][16][17] The electrochemical synthesis of ammonia at standard temperature and pressure is problematic because the nitrogen reduction reaction (NRR) competes with the hydrogen evolution reaction due to the high covalent bond enthalpy of nitrogen (N≡N bond, 941 kJ mol −1 ), which is energetically unfavored relative to the formation of protons or hydronium ion favoring hydrogen reduction and adsorption instead of nitrogen adsorption and activation on the active catalyst sites. 18 The electrochemical synthesis of ammonia is confirmed to the laboratory and not commercialized except for Japan Gasoline Co. (JGC), which produces hydrogen through water electrolysis, and operates a Haber-Boschtype reaction ( Figure 1A). 19 A variant is to utilize a solidstate electrolyte operating at a higher temperature (>400°C) in a state electrolysis cell (solid oxide electrolysis cell [SOEC]), which Marnellos and Stoukides first reported ammonia production using SOEC with protonconducting electrolytes from N 2 and H 2 using nitrogenase-based synthesis as an inspiration. 20 In the electrochemical reduction, the proton reforms hydrogen, resulting in low Faradaic efficiencies (FEs). The solubility of neutral nitrogen is low in aqueous electrolytes and partitions between the gas and liquid phases. 21 The competing oxygen from the electrolyte preferentially adsorbs into the catalysts, 22 limiting the upper FE. A solid-state oxygen ionic conductor can channel nitrogen and water to the cathode driven by higher temperatures but at standard pressure to minimize these effects. The covalent bonds of H 2 O break on the cathode to generate protons (H + ) and oxygen ions (O 2− ). The H + reacted with activated nitrogen to form ammonia, while O 2− transported to the anode and formed pure oxygen as a byproduct, adding value to the synthetic process. The advantage is that hydrogen atoms are directed from water molecules without going through molecular hydrogen, improving the kinetics, and lowering the energy demand. Like the Haber-Bosch process, yield is increased with voltage instead of pressure to promote bond fission and increased synthesis rate. Skodra and Stoukides showed this approach, 23 who reported using electrochemical synthesis with yttria-stabilized zirconia as an electrolyte in 2009, with an ammonia production rate of 1.50 × 10 −13 mol s −1 cm −2 . In other studies, Pt| gadolinium-doped ceria (GDC)|Pt button cell was used for electrochemical synthesis, and the highest ammonia production rate of 3.67 × 10 −11 mol s −1 cm −2 at 600°C was reported. 24 However, the reported ammonia synthesis rate was still lower than 4.35-8.75 × 10 −7 mol s −1 cm −2 which is a benchmark for industrial production. 25 In this contribution, we present ammonia synthesis using nitrogen and water, and lanthanum strontium chromium ferrite (La 0.33 Sr 0.67 Cr 0.33 Fe 0.67 O 3−δ , LSCrF) modified with ruthenium (Ru) ( Figure 1B,C) electrocatalyst. The SOEC device ( Figure 1D) exhibited an ammonia production rate r ( ) NH 3 of 2.37 × 10 −10 mol s −1 cm −2 at 550°C and 1.6 V, significantly higher than reported elsewhere in the literatures. Single cells were assembled for SOEC using H 2 O and N 2 as reactants (the insert in Figure 1D), yielding an ammonia production rate using Ru-modified LSCrF-GDC reached 4.73 × 10 −10 mol s −1 cm −2 . The comparison with other configurations ( Figure 1E) indicated that the exsolution of Ru from the LSCrF resulted in a significant increase in the reaction rate of ammonia synthesis.

| STRUCTURAL CHARACTERIZATION
The X-ray powder diffraction (XRD) patterns of synthesized LSCrF, Ru-doped LSCrF (LSCrFRu), and Ru nanoparticles (NPs) modified LSCrFRu (ex-LSCrFRu 1.0 h) powders are shown in Figure 2A. The major phase of the three materials is well indexed with standard LaCrO 3 (PDF 75-0288) with the Pm3m (221) space group (a = 3.9272 Å, α = 90.000°, and cell volume = 60.567 Å 3 ). For ex-LSCrFRu 1.0 h, SrFeO 3−x phase disappeared, but a trace of SrCrO 4 (PDF 73-1082, P21/c (14), a = 6.8417 Å, b = 7.5039 Å, c = 8.7292 Å, α = γ = 90.000°; β = 126.523°, and cell volume = 360.142 Å 3 ) was observed as impurity. The partial exsolution of Ru atoms caused this change after reduction at 800°C for 1.0 h under 10-vol% H 2 /Ar atmosphere. However, no peaks of Ru metal were found using the XRD due to its trace amount. An enlargement of the XRD peak at 32.44°(one of the material's major peaks) shows different degrees of splitting ( Figure 2B). This splitting indicates the existence of lattice distortion caused by Ru doping and exsolution. Two trends in the opposite direction were observed. First, the principle peak position of the crystallographic plane of (110) also shifted to a smaller diffraction angle upon Ru-doping at the B-site. Second, the diffraction angle reverted slightly to a high angle through in situ exsolution of Ru upon reduction. In this manner, the Ru-doping resulted in an initial increase in lattice parameter (a), followed by a decrease after reduction. The lattice parameters and the crystallite size calculated using Debye-Scherrer (D-S) equation (Equation S1) are also given in Table S1. Using the dimensions of the perovskite crystal cells shows that Ru is partly replaced by Fe ion at the B-site because of a similar ion radius between Ru 4+ (0.620 Å, six-coordinate) and Cr 3+ (0.615 Å)/Fe 3+ (0.645 Å, six-coordinate, highspin) in doped LSCrF (LSCrFRu), and after reduction, the Ru ion move to the surface of perovskite (ex-LSCrFRu) and reduced to Ru metal ( Figure 2C).
The high-resolution transmission electron microscope images of ex-LSCrFRu particles ( Figure 3A,B) show lattice fringes with distances of 3.8571 and 2.8333 Å, which are well indexed with the spacing of (100) and (110) planes of LaCrO 3 perovskite phase, accordant with interplanar crystal spacing interpreted using XRD. Ru NPs with a size of 4.4 ± 0.5 nm are semiembedded on the surface of the LSCrFRu matrix due to their atomic radii difference ( Figure 3B,C). As shown in Figure 3B, the lattice fringe of the NPs shows a 2.1429 Å distance, which is well indexed with the (002) plane of Ru metal (PDF 88-1734).
A linear scan of Ru-modified LSrCrFe using energy dispersive spectroscopy was collected to analyze the composition of embedded Ru NPs (from points 1 to 2 in Figure 3C), confirming the formation of metallic Ru NPs. The electrons from the 3d-orbital in Ru returned to the 3p-orbital, releasing energy corresponding to the La 1 . Figure 3E shows an enlargement of the black circle ( Figure 3B), where the contact angle (°) between Ru NPs and the LSCrFRu matrix was measured to be 57°. Assuming an atomic radius (r) of 89 pm for Ru 0 gives a vertical rise (b) of 64 pm, estimated using Equation (S2) and schematically shown in Figure 3F. The contact angle representing the surface anchorage shows that the exsolution of Ru NPs from doped LSCrFRu increases catalytic surface area, consequently promoting the ammonia production rate.

| 2297
The above results confirmed the formation of immobilized metallic Ru NPs exsoluted on the surface of LSCrFRu perovskite during the reduction process with defined crystallinity extended the length of the triplephase boundary (TPB). The exsoluted NPs provide more actives for the NRR process on the cathode surface. The exsolution-assisted immobilization of Ru on the surface of the LSCrFRu matrix was due to the donation of (2p) electron lone pairs from LaCrO 3 oxygen to the (3d) orbitals in Ru metal, which also limited Ru aggregation. This interpretation was confirmed using X-ray photoelectron spectroscopy, where the full spectrum ( Figure 4A and Table S2) shows the principle emission profile of Fe and Cr elements from 2p electron excitation, O element from 1s electron excitation in both LSCrF and ex-LSCrFRu 1.0 h. The atom ratio of Ru 4+ and Ru 0 estimated from the relative areas of fitting peaks were 91.12% and 8.88%, respectively. This result also confirmed the exsolution of Ru NPs on the surface of LSCrF perovskite ( Figure 4B). The detection of O emission is correlated with lattice oxygen and adsorbed oxygen's 1s excitation ( Figure S1a,S1b), with the atom ratio of O lat and O ads, changed from 29.1%ː70.9% to 27.5%ː72.5%, indicating that after H 2 reduction, the lattice oxygen pool decreased. The detection of Fe emission is correlated with Fe─O bonding due to 2p excitation ( Figure S1c,S1d).
The peak area of Fe 3+ 2p 3/2 in LSCrF and ex-LSCrFRu 1.0 h are 41.39% and 36.36%, while the peak area of Fe 2+ 2p 3/2 in LSCrF and ex-LSCrFRu 1.0 h are 29.82% and 32.82%, respectively. The result shows that a small amount of Fe 3+ was reduced to Fe 2+ during the 10-vol % H 2 /Ar treatment. The detection of Cr emission is correlated with Cr─O bonding due to 2p excitation ( Figure S1e,S1f). According to the peak area, the atom ratio of Cr 3+ and Cr 6+ was calculated to be 56.42% and 43.58% in LSCrF, 54.10%, and 45.90% in ex-LSCrFRu 1.0 h with no observable change in lattice coordination. The changes in oxygen vacancy were also confirmed using iodine titration, which showed a stoichiometric oxygen content decrease from 2.87 to 2.83 after the exsolution of Ru, corresponding to the increase of oxygen vacancy from 0.13 to 0.17. In contrast, the average B-site ion valence (v 0 ) decreased from 3.40 to 3.32 (for LSCrF and ex-LSCrFRu 1.0 h, Table S3). The decrease in the lattice oxygen was attributed to the exsolution of the Ru element from the B-site in perovskite crystalline structure, accompanied by B-site defects. Charge balance in the perovskite was maintained by the generation of oxygen vacancies, confirmed by thermogravimetric analysis under 10-vol% H 2 /Ar ( Figure S2). The oxygen vacancies on the perovskite lattice facilitated the transportation of O 2− and supplied reactive sites for N 2 adsorption and activation, collectively improving the ammonia production rate under the same experimental conditions, confirmed by XRD ( Figure S3) and the porous structure of the electrode microscopically ( Figure S4).

| ELECTROCHEMICAL SYNTHESIS OF AMMONIA
The ammonia production rate using ex-LSCrFRu-GDC 1.0 h was optimal under 1.6 V and compared with the LSCrF-GDC electrode ( Figure 5). The impedance spectra and their Arrhenius profiles of LSCrF-GDC and four ex-LSCrFRu-GDC reduced at 600°C are shown in Figures 5A and S5a. The intercept at the real axis for 1.0 h reduced catalyst in 10-vol% H 2 /Ar exhibited the smallest R p value of 22.97 Ω cm 2 . The ex-LSCrFRu-GDC electrode reduced for 1.0 h showed the lowest activation energy value of 1.534 eV. The lower activation energy corresponds to the more facile migration of oxygen ions, conducive to faster oxygen reduction reaction (ORR) kinetics. 22 The electrochemical impedance spectroscopy profiles were fitted to an equivalent circuit shown in the inset of Figure S5b. The fitted R n values of LSCrF-GDC and ex-LSCrFRu-GDC 1.0 h ( Figure S5c) indicated that Ru exsolution decreased resistance as anticipated. The low-frequency resistance (R 3 ) is the rate-controlling step of the ORR for ex-LSCrFRu-GDC 1.0 h and its decrease is attributed to the increased interfaces of exsolved Ru NPs embedded in the perovskite LSCrFRu matrix. As a result, these embedded Ru NPs on the surface extend the TPB region for O 2− incorporation into the lattice of the perovskite matrix. The exsolutions of Ru to the lattice surface promoted rapid O 2− transportation. The polarization curve of the single-cell with ex-LSCrFRu-GDC as the cathode showed a smaller semicircle polarization than that of the LSCrF-GDC electrode at 600°C ( Figure 5B). The stability of current density was measured under an externally applied voltage range of 1.4-2.0 V and a temperature range of 450-650°C using LSCrF-GDC electrodes ( Figure S6a). With the applied voltage increased, the current density rose, resulting in increased NH 3 yield at each selected temperature. The temperature increase to 550°C also enhances the current density due to the rapid thermal motion.
The single electrolysis cells were assembled to evaluate NH 3 production yield under different conditions. The ex-LSCrFRu-GDC cathodes showed maximized ammonia production rate at 1.6 V and 550°C ( Figures 5C and S6b). The maximum value of 4.73 × 10 −10 mol s −1 cm −2 doubled that obtained using the LSCrF-GDC cathode. The production curves ( Figure S6b) showed a volcanic shape as a function of reaction temperature, achieving the optimal yield at 550°C. This observation is attributed to the high thermal motion and the decomposition of NH 3 above 550°C. The current density increased exponentially with the reaction temperature ( Figure S6c), as evidenced in Figure S5a. The ex-LSCrFRu-GDC electrode displayed higher current density than the LSCrFRu-GDC electrode under the same condition ( Figure S6c), indicating the exsolution of Ru enhanced the transport of the charge carriers.
The applied voltage was also investigated to determine FE changes ( Figures 5D and S7). The FE value ( Figure 5D) of the NH 3 production using an ex-LSCrFRu-GDC catalyst is superior to that using an LSCrF-GDC cathode at 500°C and 550°C due to the higher ammonia production rate. Above 550°C, the rate decreased due to the decomposition of ammonia to N 2 and H 2 . 26 (Above 600°C, no significant difference was found in production rates and FE values for both ex-LSCrFRu-GDC 1.0 h and LSCrF-GDC electrodes, although the FE maxima shift to lower temperatures at higher overpotential ( Figure S7). Compared with other literature reports using H 2 O and N 2 as reactants, the ammonia production rate in our work is at least one magnitude higher, as shown in Figure 1f and Table S4. 23,24,[27][28][29] At higher temperatures, higher current density supplied more electrons for ammonia production (Equations S3-S7), increasing NH 3 yield. The increase in ammonia yield is in part due to the kinetic energy of the reactants (N 2 and H 2 O steam) increasing with the higher applied voltage ( Figure S6c). Since the ammonia formation reaction (Equation S7) is an exothermic reaction, the equilibrium of ammonia formation shifted to the left (in the direction of endothermic reaction) when the temperature increases. 30 Above 500°C, the decomposition of NH 3 to N 2 and H 2 occurred (Equation S8), resulting in a decrease in ammonia yield ( Figure S6). 26 The yield change was attributed to the endothermic reaction profile at a higher temperature and the competition of hydrogen production from water splitting that collectively decreased the NH 3 yield. 29 These factors collectively determined the maximum NH 3 yield at 550°C, as 3.83 × 10 −10 mol s −1 cm −2 at 2 V and 2.23 × 10 −10 mol s −1 cm −2 at 1.6 V ( Figure S6b). The applied voltage was controlled at 1.6 V to retain the electrode's crystalline structure. The FE value of NH 3 yield ( Figure S7) also showed a volcanic-shaped relationship with temperature. The FE increased with higher applied voltage. It was found that the maximum FE value occurred at 500°C under 1.8 and 2 V. In comparison, at 550°C under a lower voltage of 1.4 and 1.6 V, the rate was lower at the lower overpotential, which increased at 1.6 V (Figures S6c and S7a,S7b for FE). Our results support ammonia synthesis at low temperatures and high voltage to minimize the extent of the endothermic reaction.

| MECHANISM OF AMMONIA SYNTHESIS
The higher intrinsic catalytic activity due to in situ exsolved Ru incorporation was attributed to the spillover of O or O 2− species migrating on the metal surface through the TPB. 31 These effects are anticipated to be caused by the perovskite lattice and a reduction of total surface acidity by Ru ( Figure S8), leading to the chemisorption of NH 3 on weak or medium acid lattice surface sites (Table S5). The exsolution of Ru extended the TPB length and accessible sites. More active sites allowed for the N 2 adsorption at the B-site of the perovskite and activation on the desorption surface of the cathode. 32,33 At the same time, the Ru element's exsolution formed numerous oxygen vacancies in ex-LSCrFRu 1.0 h, which can also enhance the adsorption and activation of N 2 , further improving the ammonia synthesis. 34 In addition, more oxygen vacancy generation also facilitated the ORR process by generating more O 2− through an H 2 O splitting reaction (Equation S4) on the surface of ex-LSCrFRu 1.0 h than on the LSCrF catalyst ( Figure S9). Correspondingly, more H + was provided for hydrogenation-reaction, improving the ammonia synthesis.
An active solid ionic conductor was synthesized through exsolution for direct electrolytic ammonia synthesis at ambient pressure. The data indicated N 2 activation at the ex-situated Ru sites, leading to coupling with hydrogen from water and facilitating ammonia synthesis. The ammonia synthesis rate reached the maximum at 550°C under each applied voltage. With Ru modification, the ammonia synthesis rate was 4.73 × 10 −10 mol s −1 cm −2 at 550°C under 1.6 V, nearly double that using LSCrF-GDC as a cathode. The exsolution works by intrinsically enhancing Ru's catalyticity coupled with oxygen vacancies' generation by forming O 2− . The anion was generated from H 2 O splitting with H + , which participated in the hydrogenation step in the NRR process. On the basis of the above reasons, the ammonia synthesis rate has increased significantly. The research demonstrated a possible way to design an electrocatalyst for the green synthesis of ammonia with renewable electricity.