Electrocatalytic synthesis of ammonia by surface proton hopping

We accomplished efficient electrocatalytic low-temperature ammonia synthesis with the highest yield reported to date.

mm o. d.) were inserted into the reactors as electrodes. The upper electrode was set on the top of the catalyst bed. The ground electrode was set on the bottom of the catalyst bed. The catalyst bed temperature can be measured using a thermocouple. The imposed current and response voltage waves were observed using a digital phosphor oscilloscope (TCP A 300, and TDS 2001C; Tektronix Inc.).

In-situ Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS)
measurements. To elucidate the adsorbed species on catalysts with and without the electric field, insitu DRIFTS measurements were conducted using FT-IR (FT-IR6200; Jasco Corp.) with an MCT detector and a diffuse reflectance infrared Fourier transform spectroscopy reactor cell (DR-600Ai; Jasco Corp.) with the ZnSe window. For IR measurements with application of the electric field, two DRIFTS cells were used. They were made of SUS304 for catalytic reaction (for tolerance of high temperatures) and of Teflon for catalytic reaction in the electric field (to avoid short circuits in the cell), as presented in supplementary materials figure S2. The sample was 9.9 wt%Cs / 5.0 wt%Ru / SrZrO 3 . First, the sample was pre-reduced under N 2 (15 SCCM) and H 2 (45 SCCM) flow at 473 or 723 K for 2 h. Then, the reducing gas was purged for 30 min with Ar (60 SCCM). Then background (BKG) measurements were taken under Ar gas at 473 K or 648 K. Subsequently, the reactant gas (N 2 : H 2 = 15 : 45 SCCM or N 2 : Ar = 15 : 45 SCCM or 10%NH 3 /He : Ar = 1 : 59 SCCM, or H 2 /D 2 = 15 SCCM) was supplied for 30 min at most. Subsequently, the electric field was applied for about 10 min. Each spectrum was recorded at resolution of 4.0 cm −1 , over 50 scans. The imposed current was 6 or 10 mA.
Characterization of catalyst. The crystalline structure was characterized using powder X-ray diffraction (XRD, SmartLab3; Rigaku Corp.) operating at 40 kV and 40 mA with Cu-K radiation.
The XRD measurement results are presented in supplementary materials figure S9. The dispersion ratio and particle diameter of Ru catalysts are characterized using CO pulse (BEL CAT II; Bel Japan, Inc.). Before measurements, the catalyst sample was pre-treated under H 2 (50 SCCM) flow at 723 K for 30 min. After the treatment, the temperature was decreased to 323 K with He, and 10% CO was pulsed. Results of the CO pulse are presented in supplementary materials table S3. The structure of supported Ru was observed using a TEM (JEM-2100F, 200kV).
Theoretical calculations for ammonia synthesis on Ru with/without the electric field. All the calculations were conducted using the Vienna ab initio simulation package (VASP) 5.4. A corevalence effect was included with the projector-augmented wave (PAW) method [43,44]. The valence part wave functions were expanded using plane-wave basis sets with kinetic energy lower than 400 eV. For the calculation of energetics, i.e. adsorption energy, reaction energy, and activation barriers, re-parameterized Perdew-Burke-Ernzerhof (RPBE) was used as the exchange-correlation functional for spin-polarized DFT calculation [45,46]. For vibrational frequency calculations, PW91 exchangecorrelation functional was used [47]. The DFT calculations were done in a spin-polarized manner. For k-space, integration was conducted by taking 3 × 3 × 1 points. For the smearing method in the electronic state, the first-order Methfessel-Paxton method with  = 0.2 and tetrahedron method were used, respectively, for catalyst system and free-molecules (N 2 , H 2 , etc.). Both Ru(0001) and Ru (101) surfaces were modeled by four Ru atomic layers. During geometry optimization, the lower and upper two layers of Ru were fixed and relaxed, respectively. The geometry of adsorbates was fully relaxed.
The catalytic system was represented by repeated slabs, which were separated by an approximately 20 Å vacuum layer. The definitions of adsorption sites on Ru(0001) and Ru(101) are presented in supplementary materials figure S10. For a transition state search, nudged elastic band (NEB) method was used. The electric field effect was considered by adding or removing electrons from the model system. Monopole and dipole corrections were included. Vibrational frequencies of CO were obtained by diagonalizing the Hessian matrix, which was computed from a finite-difference method. The C-O stretching frequencies perpendicular to the surface were calculated.
Apparatus and IR cell for applying electric field to catalyst bed. We prepared original reactor for application of the electric field to a catalyst bed. Schematic images of reactor are portrayed in figure S6. We inserted a thermocouple into the reactor to measure the catalyst bed temperature. In addition, a hand-made Teflon cell was used to conduct DRIFTS measurements during application of the electric field to the catalyst bed. A schematic image of the Teflon cell for in-situ DRIFTS measurements is presented in figure S2.
Screening tests and activities for ammonia synthesis over 9.9wt%Cs/5.0wt%Ru/SrZrO 3 catalyst in the electric field. We chose 9.9wt%Cs/5.0wt%Ru/SrZrO 3 catalyst for detailed investigation because that catalyst showed the highest performance in our screening tests, as presented in table S4. These screening tests were conducted under the same conditions, at 473 K, 0.1 MPa, 200 mg catalyst, 6 mA current, and gas flow (N 2 = 60 SCCM, and H 2 = 180 SCCM). As Figure S13 shows, 9.9wt%Cs/5.0wt%Ru/SrZrO 3 catalyst showed a stable activity for 5 h. We conducted screening tests for catalyst support in terms of the ability of proton conduction [48][49][50]. Also, the property of not only protonic but also mixed ionic and electronic conductor is considered to be significant because electron should run through the catalyst support during the electric field application, proton hopping at the surface of the catalyst. According to the numerous researches, SrZrO 3 -based oxides show a mixed conductivity with its defect chemistry, even under dry and low temperature conditions [51][52][53]. Therefore, SrZrO 3 -based catalyst is considered to show the stable and high activity among these catalysts.
XRD patterns for each support are presented in figure S9. The crystalline structure was almost identical before and after reactions in the electric field. Using Ru catalyst, Cs is known as a promoter, which shows the electron donor effects [5]. Our tests showed that the best amount of Cs is 9.9wt% for 5.0wt%Ru-loaded catalyst. Cs dopant is positively effective to some extent because both the associative reaction and the dissociative one would proceed even when the electric field is applied to catalyst bed at low temperatures. Also, H + is partially removed from NH 4 + , which results in producing ammonia thanks to the ability of electron donor for Cs. Furthermore, the activities for ammonia synthesis with or without the electric field at various reaction temperatures, are presented in table S5.
Isotope exchange tests and the calculation for N 2 dissociative rate in the electric field. To avoid the influence of hydrogen poisoning on isotope exchange tests, we conducted isotope exchange tests without the electric field and only supplying N 2 species. Before experiments, hydrogen was purged at 723 K with Ar = 48 SCCM for 30 min. Then 28 N 2 and 30 N 2 were supplied. These results are presented in figure S1. 29 N 2 was not detected even without hydrogen supply, indicating that the N 2 dissociative rate without the electric field is very low: under the limit of detection.
However, we can calculate the N 2 dissociative rate from the detected outflow rate of 29 N 2 .
Estimating the reaction mechanism as Langmuir-Hinshelwood mechanism, steady state, and the same zero-point motion energy for 14 N and 15 N [34, 35], the following equations (1) -(3) are obtained.
Balance equation for N species flow; Equation for N 2 outflow; In those equations, V stands for the total flow rate, F signifies flow of each species, and f represents the molar fraction of isotopic species. Subscripts 28 , 29 , 14, 15 respectively denote 28 N 2 , 29 N 2 , 14 N, and 15 N. Also s signifies the Ru surface. Subscript 0 denotes an input value (see also figure S11). From the experiment, we obtained r NH3 , F 28 , (F 28 ) 0 , (f 28 ) 0 , and F 29 . Then V in , V out , and f s14 were calculated.
Results of analyses showed that the N 2 dissociative rate, V in or V out , can be calculated as about 36,000 μmol g -cat -1 h -1 and show f s14 as presented in tables S6 and S7.
Influence of metal precursor on ammonia synthesis in the electric field. Our analysis of IR measurements revealed that NH 4 + Clwas produced in the electric field. Actually, Cl is regarded as derived from the metal precursor: RuCl 3 (3H 2 O). Therefore, to investigate the influence of Cl on activity, we prepared Cl-free catalyst using metal precursor Ru(acac) 3 . Table S8 presents activities for both catalysts: Cl precursor and acac precursor. The ammonia synthesis rate, TOF (using CO pulse results, shown in table S3), and the apparent activation energy were nearly equal for the two catalysts. Table S9 and figure S12 present results for isotope exchange tests. The N 2 dissociative rate was calculated using the same procedure for Cl precursor catalyst (table S10). These results demonstrate that both catalysts exhibited almost identical performance, and also that N 2 dissociative rate per unit of electric power was the same value. Therefore, the mechanisms for ammonia synthesis in the electric field with two reactions are regarded as the same. The Cl species is not related to the promotion of ammonia synthesis in the electric field.

Theoretical calculations for ammonia synthesis with/without the electric field on Ru.
From experimental results on the ammonia synthesis on Ru, we considered that two reaction mechanisms are plausible for ammonia synthesis: One is well known as a multistep reaction that is expressed as following elementary steps (10) -(15) in the main text [54], which is named a dissociative mechanism. The other one is designated as an associative mechanism because the association of N 2 and surface H atom takes place before N 2 dissociation, which is expressed as the following elementary steps (16) -(21). Here, the N-N bond dissociation process was assumed to take place from N 2 H species.
The difference between two mechanisms was investigated using results of theoretical calculations. First, we consider the main reaction sites for ammonia synthesis. Figure 4 and figure S14 show TEM images of Ru particle supported on SrZrO 3 , and the proposed models of for Ru particle.
The figures show that the Ru particle surface consists mainly of Ru(0001), Ru(101) and Ru (100) facets. Among them, Ru(0001), Ru(101) are exposed to the particle surface. Based on this experimentally obtained result, we inferred that the mainly exposed facets of Ru particle are Ru (0001) and Ru(101). The theoretical calculations on these surfaces were considered. The definitions of adsorption sites on Ru(0001) and Ru(101) are presented in figure S10.
Before the investigation on NH 3 synthesis, the effects of electric fields on the catalyst system were examined by making comparison between experimental and theoretical results on in-situ IR spectroscopy using CO as probe molecules. The experimental results are presented in figures S3 and S4. As shown in figures S3 and S4, the appearance of the obtained peaks changed into broad ones with the electric field application. If we use each background spectrum (BKG) which was recorded at 523, 573, and 623 K against the obtained spectrum at 473 K in an electric field, then almost the same spectra with the original one (background at 473 K) were obtained, as presented in figure S5. Also as figure   S6 shows, the obtained spectra at high temperature around 673 K without an electric field demonstrate that only gas phase of CO peaks were observed due to high temperature. These two results indicate that the change of peak appearance (broad peaks) by the electric field application is not derived from the catalyst bed heating, but from various CO adsorbates. One research group elucidated the polarization effect of catalyst for ammonia synthesis with DC current supply [22]. There might be an analogy, indicating that our Ru catalyst is polarized to some extent during the electric field application, resulting in producing various CO adsorbates (broad peaks). So we estimated the electron state of Ru surface via the peaks of CO adsorbate with/without the electric field. When using Cs-Ru catalyst, the peak assigned to linear CO around 1986 cm −1 [55,56] shifted to around 2000 cm −1 , as shown in figure   S3. Moreover, when using Ru catalyst without Cs, the peaks assigned to adsorbed CO on a hollow site around 1750 and 1890 cm −1 [57][58][59] show a clear blue-shift of about 20 -30 cm −1 , as shown in figure   S4. This blue-shift with Ru catalyst was more clearly and larger than that with Cs-Ru catalyst. These results implied the effect of Cs addition, which is regarded as an electron donor toward Ru [6-11].
However, the blue-shift of peaks derived from CO vibrational frequency was observed in both cases with/without Cs addition, especially without Cs addition only when the electric field was applied to the catalyst bed. The blue-shift of peaks assigned to adsorbed CO reflected that the electron state of Ru changed by application of the electric field. Previous studies revealed that the peaks assigned to adsorbed CO showed blue shift when Ru was oxidized to some degree [55,56]: this blue-shift was observed when the electron state of Ru became positive.
This blue-shift observed experimentally was confirmed by our theoretical calculations. Table   S2 presents the influence of Ru charges on CO adsorbed onto Ru(0001) and Ru(101) by calculations.
On Ru(0001), it is widely known that CO preferentially adsorbs on either on-top of hcp three-fold hollow sites [57][58][59]. Thus, we considered CO adsorptions on these sites. Here, the electric field was Therefore, we can conclude that the Ru surface under electric field can be expressed by introducing positive charges on or removing electrons from the system.
Next, to elucidate the detailed reaction mechanism of ammonia synthesis under electric field, we investigated the detailed energetic change of elementary reactions along the reaction path. We consider the ammonia synthesis reaction proceeds via dissociative or associative mechanism. Previous theoretical study by Garden el al. have suggested that the rate-determining step of the dissociative and associative mechanism is the N 2 dissociation and N 2 H formation steps, respectively [42].
Based on this model, we carried out geometry optimization of transition state for N 2 dissociation and N 2 H formation reactions on Ru. Both (0001) and (101)