Favorable Role of the Metal-Support Perimeter Region in Electrochemical NH3 Synthesis: A Density Functional Theory Study on Ru/BaCeO3.

The catalytic electrochemical synthesis of NH3 on Ru/BaCeO3 was investigated using density functional theory. The competition between NH3 formation and the hydrogen evolution reaction (HER) is a key for a high NH3 formation rate. Our calculations show that H adsorbs more strongly than N2 at the Ru particle moiety, while the adsorption of N2 is stronger than the H adsorption at the Ru/BaCeO3 perimeter, a model for the triple-phase boundary that is proposed to be an active site by experimental studies. This indicates that, while the HER is more favorable at the Ru particle moiety, it should be suppressed at the Ru/BaCeO3 perimeter. We also calculated the Gibbs free energy changes along the NH3 formation and found that the N2H formation, the NHNH2 formation, and the NH3 formation steps have a relatively large Gibbs energy change. Therefore, these are possible candidates for the potential-determining step. The calculated equilibrium potential (U = -0.70 V, vs RHE) is in reasonable agreement with experiments. We also evaluated the reaction energy (ΔE) and the activation barrier (E a) of the N2H formation at several sites. ΔE and E a were high at the Ru particle moiety (ΔE = 1.18 eV and E a = 1.38 eV) but became low (ΔE = 0.32 eV and E a = 1.31 eV) at the Ru/BaCeO3 perimeter. These provide the atomic-scale mechanism how the proton conduction in BaCeO3 assists the electrochemical NH3 synthesis.

Introduction

Recently, ammonia (NH3) has attracted considerable attention as a hydrogen (H2) energy carrier for several reasons. First, NH3 is much easier to liquefy and transport than H2. Second, NH3 has a higher volumetric H2 density (120.3 kg-H2·m–3) than liquid H2 (70.9 kg-H2·m–3) or toluene/methylcyclohexane (47.1 kg-H2·m–3).[1] Despite these advantages of NH3 as a H2 carrier, the Haber–Bosch process[2] remains the main synthetic route for producing NH3 from N2. However, the main hydrogen source for this process is currently the hydrogen gas formed from natural gas. This requires large amounts of energy; thus the NH3 formation without using hydrogen from natural gas is highly desirable. For this purpose, electrochemical NH3 synthesis is a promising approach because it utilizes H2O as a hydrogen source.[3] Electrochemical NH3 synthesis could be a key technology for a carbon-neutral society.

One promising approach for electrochemical NH3 synthesis is to employ a fuel cell with a solid electrolyte. Many studies have reported using a solid electrolyte for NH3 synthesis at the gas–solid interface.[4,5] This device consists of an anode, a cathode, and a solid electrolyte. The anode and cathode reactions areandrespectively, and the net reaction isWater dissociation on the anode generates protons, which are transported to the cathode through the solid electrolyte and combine with electrons and N2 there to produce NH3. The NH3 formation at the cathode is known to determine the overall reaction rate and yield, and extensive efforts have been devoted to finding effective catalysts for the cathode reaction. For example, several research groups have investigated metal systems such as Ru and Ni. Ru is well known as a suitable catalyst for NH3 synthesis because it exhibits a high NH3 formation rate.[6] However, the electrochemical NH3 formation rates reported are still low, such as 3.0 × 10–13 mol·s–1·cm–2 on Ru/MgO, as observed by Skodra and Stoukides.[7]

To further enhance the NH3 formation rate for practical purposes, a detailed understanding of the reaction mechanism is necessary. Several mechanistic studies based on kinetic measurements or theoretical calculations have been performed.[3] Currently, there are two widely accepted pathways for NH3 synthesis: the dissociative mechanism (in which N–N bond dissociation occurs first) and the associative mechanism (in which NH (x = 1–3) species form before N–N bond dissociation).[8−11] The Haber–Bosch process is considered to occur through the dissociative mechanism, while many researchers believe that electrochemical NH3 synthesis occurs through an associative mechanism because of its milder reaction conditions.[9,12] Another important mechanism is the Mars–van Krevelen mechanism, in which lattice N atoms are used for NH3 formation; several research groups have investigated electrochemical NH3 synthesis via this mechanism using a metal nitride cathode.[3,13]

Another key factor in the electrochemical NH3 synthesis is its competition with the hydrogen evolution reaction (HER).[5,14] This reaction is expressed asH adsorption on the catalyst surface or active site competes with N2 adsorption, and when a negative potential is applied, the HER becomes easier than NH3 formation. This is unfavorable for the NH3 formation because the active sites and the electronic current are consumed by the HER. Some catalyst systems with high N2 dissociation rates (e.g., Ru and Ni) suffer from low NH3 formation rates because of the HER; several groups previously analyzed the competition of the HER and NH3 synthesis and have shown that overcoming the HER is necessary condition to have a high NH3 formation rate.[12,15]

As stated above, many mechanistic insights into electrochemical NH3 synthesis have been obtained. However, several important features remain uninvestigated, such as the metal–support interaction, which is known to play an important role in catalysis.[16] Indeed, the triple-phase boundary (TPB) is considered to be the active site for electrochemical NH3 synthesis in the gas phase.[17] Therefore, identifying the detailed mechanism at the TPB is of particular importance. One of the authors of the present study found that, in the Ru/BaCeO3 system, a smaller Ru particle size is favorable for electrochemical NH3 synthesis because of the increased Ru/BaCeO3 perimeter region, which has an NH3 formation rate of 1.1 × 10–11 mol·s–1·cm–2.[17] Recently, we proposed the formation of an effective double layer during NH3 electrosynthesis with Fe/BaCeO3, which can be considered as an extension of the TPB active site model.[18] Several researchers have investigated the metal–support interaction in the electrochemical reaction and noted some similarities between this interaction and the electrochemical promotion effect.[19] An understanding of the complex relationship between these two factors is important for enhancing the electrochemical catalysis process. However, there is still much to be learned on this issue.

In this study, we theoretically investigated electrochemical NH3 synthesis using a Ru/BaCeO3 system. To the best of our knowledge, this is the first computational analysis of the metal–support interaction during electrochemical NH3 synthesis. Our focus was the effect of the metal–support interaction on the mechanistic details, especially on the competition between NH3 formation and the HER and the energetics of NH3 formation. First-principles density functional theory (DFT) calculations were employed to provide a reliable description of the thermodynamic and kinetic properties. We first examined the competitive adsorption of H and N2 on the Ru surface to identify the available active sites for NH3 synthesis. Then, we investigated NH3 formation by considering the free energy profiles under an applied electric potential. Based on the energetic profile of H and N2 adsorption on the active sites, the availability of the sites, that is, competition between the HER and the NH3 formation is discussed. The calculated free energy profile is compared with experimental results. Finally, we evaluated the activation barrier (Ea) of the N2H formation step, identified as one of the key steps of electrochemical NH3 formation from the analysis of the free energy profile.

Method

Reaction Model

For electrochemical NH3 synthesis, two pathways have been discussed, namely, the dissociative and associative mechanisms.[10,11,20] With Ru catalysts, it is widely accepted that the dissociative mechanism mainly occurs at stepped surfaces.[8,21] However, the associative mechanism is more plausible for the current purpose from the following reasons: (1) considering the structure of the perimeter moiety, stepped surfaces are unlikely to be formed at the Ru/BaCeO3 perimeter, and (2) NH3 formation at the Ru/BaCeO3 perimeter occurs under proton-rich conditions because a proton is always supplied through BaCeO3. Note that the NH3 formation on Ru nanoparticles (i.e., nonperimeter sites) can be modeled using a conventional Ru slab model. As such models have been extensively studied, we did not repeat the calculations.[11,12]

In addition to differences in the order of N–H bond formation and N2 dissociation, there are two possibilities for the H addition to N or N2: the Tafel mechanism and the Heyrovsky mechanism. In the former, the N–H bond is formed between N and H atoms both adsorbed on the metal surface, while in the latter, the added H comes from a medium such as a gas, solvent, or solid electrolyte. Skúlason et al. compared these two mechanisms and concluded that the Heyrovsky mechanism is favorable in terms of the reaction energy.[12] Based on this, we considered the associative Heyrovsky mechanism, which involves the following elementary steps (the active site is denoted by an asterisk (*)):

For the associative Heyrovsky mechanism, Li et al. proposed two possible pathways depending on the position of H addition to the N2H intermediate.[22] These pathways are denoted as the alternating and distal pathways in Figure .

Figure 1

Schematic reaction mechanisms for associative NH3 synthesis via distal and alternating pathways. The shaded area denotes the Ru surface. The alternating pathway has two further possibilities shown by the black and light gray arrows.

The Gibbs free energy change for NH3 electrosynthesis was analyzed using the approach proposed by Nørskov et al.[12,23] It was calculated as follows:where ΔEDFT is the reaction energy calculated using DFT; ΔEZPE and ΔS are the changes in the zero-point energy (ZPE) and entropy along the reaction step, respectively. U is the externally applied electric potential that shifts the Gibbs energy of the reaction intermediate by −neU (n = number of electrons involved). The ZPE and the entropies of gaseous molecules (N2, H2, and NH3) were obtained from experimental data and are listed in Table S1.[24] To calculate the chemical potential of the proton–electron pair (H+ + e–), we employed the computational hydrogen electrode model.[23] In this model, the Gibbs energy G(H+ + e–) is equivalent to half of the Gibbs energy of gaseous hydrogen (1/2G(H2)) under standard conditions (pH = 0, 298.15 K, 1 atm) with no external potential.

Computational Details

We employed the DFT + U method (DFT plus Hubbard-U parameter) for all DFT calculations.[25] The Perdew–Burke–Ernzerhof functional was used for the exchange-correlation functional.[26] The effective Hubbard-U parameter, that is, U – J was set to 6.0 eV, which was used for the Ce 4f electrons as in a previous DFT + U study on BaCeO3.[27] We examined the dependence of the results on the Hubbard-U parameter by using the N2 adsorption energy on Ru/BaCeO3 as a benchmark (Figure S1). When the U parameter was varied within 0.0–8.0 eV, the N2 adsorption energy ranged from −0.47 to −0.51 eV. This indicates that the U parameter has only weak effects on the Gibbs free energy profiles.

We performed geometry optimization by fixing the lower two-thirds of the BaCeO3 structure. The Ru moiety and the adsorbates were fully relaxed. Figure displays details of the Ru/BaCeO3 unit cell model, which contained 284 atoms, with 13 atomic layers of BaCeO3 and 4 of Ru. No spatial symmetry was imposed in the calculations. To evaluate the ZPE, a vibrational analysis was performed using a finite difference of 0.015 Å and allowing only the adsorbate molecules to move. Unit cell optimization was performed for the Ru/BaCeO3 system (without an adsorbate), and the optimized unit cell parameters were used for the subsequent calculations. For the surface energy calculations, another unit cell optimization was performed. For a selected case, we performed a molecular dynamics (MD) simulation with an NVT ensemble to search for a stable adsorbate structure. A Nose–Hoover thermostat was used there, the temperature was controlled at 500 K, and the simulation was carried out for 1 ps with a time step of 1 fs.

Figure 2

Top view and two-side views of the Ru/BaCeO3 rod model used in the calculations. The lower two-thirds of BaCeO3 were fixed during geometry optimization.

The core electrons were represented using the projector-augmented wave method.[28] The valence electrons were expanded by the plane wave basis set up to a cutoff energy (Ecut) of 400 eV. The electron occupation near the Fermi level was determined using the first-order Methfessel–Paxton scheme with σ = 0.1. The convergence criteria for the electronic state and geometry optimization calculations were set to 1.0 × 10–5 and 0.03 eV·Å–1 in energy and force, respectively. Transition state (TS) search was done with the climbing image nudged elastic band (CINEB) and the dimer methods.[29] The vibrational analysis of the TSs confirmed that they were the first-order saddle point in the potential energy surface. Reciprocal space integration was performed with the k-point placed using the Monkhorst–Pack scheme. The k-point mesh was set to 3 × 3 × 1 for the surface calculations and to 9 × 9 × 9 for the bulk material calculations. Gamma-point sampling (1 × 1 × 1) was used in the MD and CINEB calculations. A vacuum layer of ∼20 Å was introduced between the slabs. Bader charge analysis was used to investigate the electronic properties.[30] Dipole correction in the z-direction was applied in all calculations except when calculating the bulk material and isolated molecules. All DFT calculations were performed using the Vienna ab initio simulation package version 5.4.[31]

Results and Discussion

Determination of the Most Stable Surface Termination

Herein, we employed perovskite BaCeO3 as the proton-conducting electrolyte. Although a previous experimental study used yttrium-doped BaCeO3 as the electrolyte,[17] the doping level was moderate (BaCe0.9Y0.1O3) and was therefore not expected to change the reaction mechanism of NH3 synthesis. The orthorhombic unit cell structure of BaCeO3 was obtained from the Inorganic Crystal Structure Database (ICSD ID = 188637). First, we need to identify the most stable surface of BaCeO3 among several candidates, because the surface should be exposed during electrochemical NH3 synthesis. Previously, Shishkin and Ziegler employed the DFT method to compare the stabilities of the (100), (110), and (111) surfaces of BaCeO3 and found (100) to be the most stable.[32] Based on their result, we considered the (100) surface of BaCeO3 in this study.

We used a 2 × 2 × 2 supercell to model the BaCeO3 surface. The surface energy (Esurf) was evaluated as follows:where A is the surface area calculated from the optimized unit cell, Eslab is the total energy of the slab, Ebulk is the total energy of bulk BaCeO3, N = nBa, slab/nBa, bulk is the number of BaCeO3 unit cells, and nX, slab and nX, bulk are the numbers of atom X in the slab and bulk models, respectively. μX is the chemical potential of atom X.[32,33] The chemical potential of an oxygen atom (μO) is calculated fromwhere EO is the total energy of O atom, and BDEO and ZPE are the bond dissociation energy and zero-point energy of O2, respectively.[34] We employed the experimentally measured values of BDEO = 4.89 eV and ZPE = 0.10 eV.[24] For the total energy calculation of O atom, DFT calculations were performed using B3LYP as the exchange-correlation functional.

The chemical potentials of Ce and Ba were evaluated aswhere the references for Ce and Ba were bulk CeO2 and BaO, respectively. The total energies (ECeO2 and EBaO) were evaluated using DFT. The initial structures of CeO2 and BaO were obtained from the ICSD (ID = 88759 and 616005, respectively). The calculated chemical potentials of H, O, Ce, and Ba are listed in Table S2. Using these quantities, we calculated the surface energies with four different terminations: CeO2 termination, CeO2–O termination, BaO termination, and BaO–O termination (see Figure S2 for their surface structures). According to the calculated Esurf values in Table , BaO termination is the most stable. Based on this result, we constructed the Ru/BaCeO3 rod model to represent the Ru/BaCeO3 perimeter. We placed a Ru rod consisting of 80 atoms on this surface. The resultant structure is shown in Figure . This model is used in the NH3 synthesis at the Ru/BaCeO3 perimeter, which will be discussed in the following sections.

Table 1

Surface Energies (Esurf) of BaCeO3(100) Surfaces with Different Terminations, as Calculated Using DFT

	CeO2	CeO2–O	BaO	BaO–O
surface energy (J·m–2)	1.38	4.75	1.03	4.34

Competitive Adsorption of H and N2

As discussed in Introduction, one reason for a low NH3 formation rate is competition with the HER. In this subsection, we investigated the adsorption of H and N2 on Ru/BaCeO3. We consider the adsorption of H and N2 to occur on the several sites on the Ru rod (top, edge, side) and the Ru/BaCeO3 perimeter sites. The top part of the Ru rod has a surface structure similar to that of Ru(0001); therefore, we assume that this part has a reactivity similar to that of the Ru surface or nanoparticles. Figure summarizes Gibbs free energies of H and N2 adsorption (ΔGad), together with the optimized geometries. First, we discuss N2 and H adsorption on the top part of the Ru rod. At this location, N2 undergoes atop adsorption, while H adsorption is exergonic at both the fcc and hcp threefold hollow sites. Notably, N2 adsorption on the top part is weak, as indicated by the positive ΔGad value (0.08 eV); this is in agreement with the previously reported tendency on Ru(0001).[12] On the other hand, H adsorption is strong (ΔGad = −0.36 eV) on the Ru particle, suppressing NH3 formation at these locations through serious H poisoning. A similar tendency was observed for H and N2 adsorption on the upper edge part. Among the Ru edge or side part of the rod, the H adsorption is most strong at the edge site with bridge type adsorption (ΔGad = −0.59 eV). The H adsorption on this site is stronger than the H adsorption on the Ru top part. N2 adsorption here is much weaker than the H adsorption (ΔGad = −0.29 eV), although still stronger than that at the Ru top part.

Figure 3

Optimized structures (top view) of H- and N2-adsorbed Ru/BaCeO3 systems and their adsorption energies. A side view is also provided for N2 adsorption on the Ru/BaCeO3 perimeter site.

Next, we consider N2 and H adsorption on the Ru/BaCeO3 perimeter. Here, H adsorption is the strongest at the bridge site (bridge(A) in Figure ), as its ΔGad is −0.79 eV; it is much stronger than that on the Ru top or edge parts. The strongest N2 adsorption on Ru/BaCeO3 occurs at the atop(A) site with side-on adsorption mode, as its ΔGad is −1.02 eV. In this configuration, two Ru atoms at the perimeter are used, where each N atom binds different Ru atoms as shown in the top view (Figure ). This adsorption mode is close to the enzymatic configuration (N*–N*), which is seen in the N2 adsorption on the enzyme nitrogenaze.[35] Another mode of N2 adsorption, atop adsorption, is also strong as its ΔGad is −0.64 eV. This is also much stronger than the N2 adsorption on the Ru top and edge parts.

Thus, these results show that the N2 adsorption at the Ru/BaCeO3 perimeter is stronger than the H adsorption, which differs from the scenario at the Ru top or edge parts. To examine the stability of this N2 adsorption site, we performed an MD simulation at T = 500 K up to 1 ps (see Figure S3 for the energy and geometry changes along the MD trajectory). Our MD calculation showed that N2 is always adsorbed on the Ru/BaCeO3 perimeter region. Thus, we can conclude that this site binds N2 with considerable strength.

To analyze the electronic properties of the Ru/BaCeO3 system, we carried out Bader charge analysis. The calculated charges of the Ru rod are summarized in Figure S4. The results show that the Ru atoms in the Ru/BaCeO3 perimeter region are more negatively charged (∼ −0.2 e) than those in other parts of the Ru rod. This could lead to stronger adsorption of H and N2 on the perimeter sites due to the possibility of stronger electrostatic interactions or back-donation to N2 π* orbitals. This interaction is expected to enhance N2 adsorption on the perimeter site, which is a unique feature not found in the Ru top or edge parts. This result strongly indicates that H2 and N2 adsorption is promoted on the TPB compared with that at other parts of the catalyst surface.

Gibbs Energy Changes during NH3 Formation

Next, we investigated the Gibbs energy changes during NH3 formation. Because the previous section concluded that N2 most strongly interacts with the Ru/BaCeO3 perimeter site, we consider it the active site for NH3 formation. Figure summarizes the Gibbs energy changes during NH3 formation on Ru/BaCeO3, considering both the distal and alternating pathways of NH3 formation. The alternating pathway also includes two possible routes, namely NHNH2* to NH* via NH3 formation and desorption, or NHNH2* to NH2NH2* formation. Therefore, we can propose three pathways for the NH3 formation, that is, (i) the distal pathway, (ii-a) the alternating pathway without NH2NH2* formation, and (ii-b) the alternating pathway with NH2NH2* formation. These pathways are shown in Figure b, where the insets display the optimized structures for each step.

Figure 4

Gibbs free energy profiles at U = 0 and −0.70 V (Gibbs free energy of the PDE, the NH3* formation step) during electrochemical NH3 synthesis via (a) distal pathway, and (b) alternating pathway. In (b), two pathways are shown; (ii-a) alternating pathway without NH2NH2* formation, and (ii-b) alternating pathway with NH2NH2* formation. The temperature was set to 298.15 K. The inset images show the optimized structures.

Based on our calculation, the elementary reactions with large ΔG values are N2H* formation from N2* (ΔG = 0.67 eV), NHNH2* formation from NHNH* in path (ii-a) (ΔG = 0.66 eV), and NH3* formation from NH2* (ΔG = 0.70 eV). Therefore, these are possible candidate for the potential-determining step (PDS, i.e., the elementary step that requires the largest ΔG). Because the N2H* formation and NH3* formation appear in all the paths, it is difficult to conclude whether the distal or alternating pathway is favorable. However, we could exclude the (ii-b) pathway because the NH2NH2* state is more thermodynamically unfavorable than the NH state. Note that the last reaction step, that is, the NH3* desorption process is endergonic in Figure , but this process becomes exergonic by 0.57 eV at T = 500 °C, which corresponds to the experimental reaction temperature.[17] Therefore, the NH3* desorption does not hinder the NH3 formation.

The theoretical overpotential can be calculated from this ΔG value by applying an external potential U to make ΔG = 0 for the PDS. Figure also shows the Gibbs energy changes under U = −0.70 V (vs RHE), which corresponds to the ΔG of NH3* formation from NH2*. One of the present authors reported in the experimental work that NH3 electrosynthesis can be initiated by applying a potential of approximately −0.3 V.[17] Thus, the present calculation exhibits semiquantitative agreement with the experimental result, although it moderately overestimates the overpotential. Our calculated overpotential value is similar to that reported by Back and Jung (−0.68 V), who considered a Ru step as the active site.[11] Therefore, the Ru/BaCeO3 perimeter site is not likely to reduce the overpotential for NH3 synthesis.

In summary, the calculated Gibbs energy profiles have shown that (i) the PDS of the electrochemical NH3 synthesis is the NH3* formation step (NH2* + H+ + e– ⇌ NH3*), while the N2H* formation step (N2* + H+ + e– ⇌ N2H*) or the NHNH2* formation step (NHNH* + H+ + e– ⇌ NHNH2*) has a similar ΔG value; thus these two are the possible candidate for the PDS, (ii) NH3 synthesis can occur at the Ru/BaCeO3 perimeter site upon applying a weak external potential, and this site is less vulnerable to H poisoning than the Ru top part is. Because we used the Ru/BaCeO3 perimeter site to model the TPB of the solid electrolyte, the DFT results indicate that the presence of the TPB is favorable for NH3 formation compared with the HER, from the thermodynamic viewpoint. In the next section, we further investigate the effect of the Ru/BaCeO3 perimeter on the kinetics of NH3 formation.

Reaction Path for the N2H Formation Process

The Gibbs energy analysis in the previous section has shown that N2H formation is a key step in the NH3 electrosynthesis, because it is involved in both distal and alternating pathways and has relatively large ΔG. Therefore, the kinetic parameters such as activation energy (Ea) or reaction energy (ΔE) for this process would be a governing factor for the NH3 formation rate. In this section, we analyze how the Ru/BaCeO3 boundary affects these parameters of the N2H formation. We investigated three pathways for N2H formation: I) at the Ru particle edge site, II) in the Ru/BaCeO3 boundary region, between the N2 molecule and H atom, both adsorbed on the Ru perimeter sites, and III) in the Ru/BaCeO3 boundary region, between a N2 molecule adsorbed on the Ru perimeter site and the H atom occupied the octahedral site in BaCeO3. We considered the path III because BaCeO3 is the proton-conducting solid electrolyte; thus the proton supply to the BaCeO3 surface is possible. The optimized structures for the reactants, TSs, and product states for the three pathways are shown in Figure , together with the corresponding ΔE and Ea values. All the TSs are confirmed to have one imaginary frequency; see Table S3 for the calculated vibrational frequencies.

Figure 5

Reactant states, TSs, and product states for N2H formation. Three reaction pathways were considered; (I) N2H formation at the Ru particle edge site, (II) in the Ru/BaCeO3 boundary region, between N2 and H both adsorbed on the Ru perimeter sites, and (III) in the Ru/BaCeO3 boundary region, between N2 adsorbed on the Ru perimeter site and H occupying the octahedral site in BaCeO3. ΔE and Ea values (including ZPEs) are shown together with the optimized structures.

Figure shows that N2H formation via pathway I has Ea = 1.38 eV and is highly endothermic (ΔE = 1.18 eV). Thus, when N2 is adsorbed on the Ru edge site, N2H formation is quite slow. In addition, our adsorption energy calculation (in Section ) shows that the H atom adsorbs more strongly than N2 at the edge site. Therefore, the active site is predominantly covered with H atoms (and thus more of the HER occurs), and N2H formation at the Ru particle is unfavorable.

Next, we consider the pathway II. As shown in Section , the adsorption energies of N2 and H at the Ru perimeter site are comparable; thus N2H formation from this site is highly probable. However, our calculations show that the reaction requires a relatively large Ea (1.60 eV) and ΔE (1.09 eV). Note that the calculated N2H formation Gibbs energy was 0.67 eV, where the coadsorption of N2 and H was not accounted. The ΔE value of the pathway II includes the coadsorption effect, and this indicates that the adsorption of H considerably stabilizes the reactant state for N2H formation. As a result, ΔE becomes larger than that in Figure . These results indicate that N2H formation at the Ru perimeter site is slow, even when N2 is strongly adsorbed on the Ru perimeter site.

Another possibility for N2H formation at the Ru/BaCeO3 boundary region is pathway III, that is, N2H formation between N2 at the Ru perimeter and the H atom from BaCeO3. Although several positions of the H atom are possible in BaCeO3, we selected the octahedral site of Figure as it is closest to the surface. Figure shows that pathway III has lower Ea (1.31 eV) and ΔE (0.32 eV) values compared with pathways I and II. Because the product state of N2H formation is the same as that of pathway II, N2H formation in the Ru/BaCeO3 boundary region is more favorable when it occurs between an N2 molecule on the Ru perimeter and H atom conducted from BaCeO3. This computational result indicates that the proton-conducting support materials such as BaCeO3 are kinetically favorable because they open a new pathway that accelerates the formation of N2H.

Conclusions

Electrochemical NH3 synthesis using a solid electrolyte is a promising approach for environmentally friendly NH3 production. Previous experimental investigations of electrochemical NH3 synthesis in the Ru/BaCeO3 demonstrated the Ru/BaCeO3 perimeter region (or three-phase boundary) to be the active site. In the present study, we used DFT calculations of a Ru rod model supported by BaCeO3 to investigate electrochemical NH3 synthesis and clarify the contribution of the metal–support interaction. First, after considering the BaCeO3(100) surface with four different terminations, the surface with the Ba termination was determined to be the most stable as it had the lowest surface energy.

We then compared the adsorption of H and N2 at several sites in the Ru/BaCeO3 system, because this is a key factor influencing the two competitive reactions, that is, NH3 formation and the HER. Because our calculated Gibbs energy of adsorption (ΔGad) shows that H adsorbed more strongly than N2 at the top part of the Ru particle, the HER is considered more favorable at that site, which corroborates previous reports. In contrast, N2 has stronger adsorption (ΔGad = −1.02 eV) than the H atom (ΔGad = −0.79 eV) at the Ru/BaCeO3 perimeter site. This suggests the weaker hydrogen poisoning on the Ru/BaCeO3 perimeter site, which is favorable for the NH3 formation.

Assuming the Ru/BaCeO3 perimeter as the active site, we also calculated the Gibbs free energy changes along NH3 formation; the associative mechanism (N2H formation prior to N–N dissociation) was assumed. Our results indicate that the PDS for NH3 formation is either the NH3* formation step (ΔG = 0.70 eV) or the N2H* formation step (ΔG = 0.67 eV). Accordingly, the theoretical equilibrium potential is U = −0.70 V, which is in reasonable agreement with the experimental value. This value is similar to the previously reported value for the pure Ru model. Considering these DFT results, the Ru/BaCeO3 perimeter site is favorable for the electrochemical NH3 synthesis because the HER is inhibited there and not because of the decrease in the overpotential of the NH3 formation. We also located the TS for the potential-determining N2H formation step. Our calculated activation barriers showed that N2H formation in the Ru/BaCeO3 boundary region is favorable, especially when N2 at the Ru perimeter site reacts with an H atom from BaCeO3. Thus, a faster rate for N2H formation is expected in the Ru/BaCeO3 perimeter region.

This investigation is the first to confirm theoretically that electrochemical NH3 synthesis is favorable at the TPB. The proton-conducting nature of BaCeO3 is beneficial for the NH3 formation because (i) it supplies the N2 activation site that has some resistance to the hydrogen poisoning, (ii) it assists the N2H formation by lowering the activation barrier, and (iii) it helps the supply of the reactant species (H atom in the present case). We consider that such a metal–support interaction might be a clue for increasing the NH3 formation rate. Our theoretical analysis strongly implies that enlarging this area is beneficial for electrochemical NH3 synthesis, from the above reasons. Based on this idea, detailed catalyst designs at the interface could enhance the reaction rate of electrochemical NH3 synthesis.