Iron particles loaded with an electron-donating material
To decrease the working temperature for ammonia synthesis over iron-based catalysts is difficult because iron has a low electron affinity to be negatively charged. We have recently reported a decrease in the working temperature of an electron-donating material in contact with Ru nanoparticles to decrease temperature for ammonia synthesis over Ru.12,16 N2 cleavage is a determinant step in ammonia synthesis and is accelerated by electron donation from an electron-donating material to adsorbed N2 molecules via transition metal d-orbitals. As a result, the working temperature for ammonia synthesis over Ru can be decreased by decreasing the working temperature of the electron-donating material. In this case, the catalyst working temperature is expected to depend largely on the electron affinity of the transition metals as the active sites because the transition metals also mediate electron donation. However, the electron affinity of iron is only 15 kJ mol-1 and is the lowest of those transition metals used to synthesize ammonia (Ru: 101; Os: 104; Co: 64; Ni: 112 kJ mol-1); an appropriate manner to combine iron with an appropriate electron-donating material must be determined. In several tested catalyst designs, only a new type of catalyst design, metallic iron particles loaded with a mixture of BaO and BaH2 (BaH2-BaO), was effective for low temperature ammonia synthesis. Conventional supported iron catalysts, where iron nanoparticles are deposited on support particles with electron-donating capability or both iron nanoparticles and an electron-donating material are deposited on support particles, could not act as catalysts for ammonia synthesis below 200 °C. For example, iron nanoparticles-deposited on BaH2-BaO (Fe/BaH2-BaO) could not act as a catalyst at 200 °C, although Ru nanoparticles deposited on BaH2-BaO (Ru/BaH2-BaO) with the same structure as Fe/BaH2-BaO could form ammonia below 200 °C16. In Fe/BaH2-BaO and Ru/BaH2-BaO, transition metal nanoparticles (2-4 nm) are deposited on BaH2-BaO phase formed on several μm CaH2 particles (see Methods)16.
Metallic iron particles loaded with BaH2-BaO (BaH2-BaO/Fe) were readily prepared by heating Ba(NO3)2-impregnated Fe2O3 particles (20‒40 nm) with CaH2 particles at 300 °C in a flow of H2. The diffraction peaks due to BaH2, BaO and metallic iron, in addition to those of CaH2 as a background, were observed in the powder X-ray diffraction (XRD) profile of the resultant BaH2-BaO/Fe (Fig. 1a); Ba(NO3)2 is decomposed into BaO on α-Fe2O3, and O2- anions in a part of BaO are replaced by H– anions of CaH2, which forms a mixture of BaO and BaH2 (BaH2-BaO) with CaO. The Fe 2p X-ray photoelectron spectroscopy (XPS) spectrum for BaH2-BaO/Fe (Supplementary Fig. S2) showed that the majority of the surface iron species on the iron particles in BaH2-BaO/Fe is zero-valent Fe. The Ba 3d5/2 peak appeared at 781 eV in the XPS spectrum for BaH2-BaO/Fe (Supplementary Fig. S3). Because the Ba 3d5/2 peak appears around 781 eV when BaH2 forms in BaO16, the Ba 3d5/2 peak in Supplementary Fig. S3 also indicates BaH2 formation in BaH2-BaO/Fe as well as the XRD features due to BaH2 (Fig. 1a). High-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and energy dispersive X-ray spectroscopy (EDX) images showed that Ba species are segregated on metallic iron particles (>10‒30 nm) in contact with CaH2 (Fig. 1b). Therefore, the preparation procedure readily leads to iron particles loaded with a mixture of BaO and BaH2, as schematically shown in Fig. 1c. We reported that BaH2-BaO exhibits strong electron-donating capability in BaH2-BaO with deposited Ru nanoparticles (Ru/BaH2-BaO)16. A part of the BaH2 in BaH2-BaO releases H atoms from H– anions as H2 molecules through transition metal surfaces, so that electrons remain in BaH2 to release H atoms (BaH2 ↔ Ba2+H–(2–x)e–x + xH↔Ba2+H–(2–x)e–x + x/2H2). The strong electron donation from the BaH2 species (Ba2+H–(2–x)e–x) to N2 adsorbed on the transition metal surfaces facilitates N2 cleavage (see below). The temperature at which H2 begins to desorb corresponds to the lowest catalyst working temperature and is expected to depend on the transition metal employed. In the case of Ru/BaH2-BaO, H2 desorbed above 140-150 °C, and ammonia formation was observed above 150 °C.16 It was confirmed that ammonia synthesis over Ru nanoparticles deposited on Ca2+F–H– (Ru/CaFH) also proceeded at 50 °C in a similar manner to that over Ru/BaH2-BaO12. H2-temperature programmed desorption (TPD) revealed that BaH2-BaO/Fe desorbs H2 above ca. 100 °C (Fig. 1d). As a result, BaH2 species (Ba2+H–(2–x)e–x) with strong electron-donating capability are formed that can donate electrons to metallic Fe at this temperature. The H2 desorption from BaH2-BaO/Fe is distinct from that of metallic iron particles without BaH2-BaO (Fe particles) obtained by heating pure α-Fe2O3 particles with CaH2. H2 desorption was observed above 200 °C from iron particles without BaH2-BaO and reached a maximum at 280 °C (Fig. 1d). The H2 desorption was due to H– in CaH2, and no ammonia formation was observed over iron particles without BaH2-BaO at temperatures below 200 °C. The electron-donating capability of Ba2+H–(2–x)e–x was estimated by density functional theory (DFT) calculations. The work functions for the (001) surfaces of Ba2+H–2 and Ba2+H–(2–x)e–x (x=1/9, Ba2+H–(2–1/9)e–1/9) were estimated to be Φ=4.2 and 2.6 eV, respectively, by DFT computation to calculate the work functions of alkaline earth metal hydrides with accuracy (Supplementary Fig. S4). Therefore, the removal of H atoms from BaH2 can form BaH2 species with high electron-donating capability, comparable to those of metallic K and Na (Φ=2.3 and 2.4 eV, respectively).
Catalytic performance of BaH2-BaO/Fe for ammonia synthesis
Figure 2a shows the catalytic performance for ammonia synthesis (100-200 °C, 0.9 MPa) over BaH2-BaO/Fe and Ru/BaH2-BaO. The results for a commercial promoted iron catalyst (promoted-Fe) that consists of Fe, K2O, Al2O3 and CaO, MgO loaded with Ru nanoparticles and Cs oxide species (Cs-Ru/MgO), and Ru-nanoparticles deposited on [Ca24Al28O64]4+(e–)4 (Ru/C12A7:e–) as benchmark catalysts are also shown. Supplementary Table S1 summarizes the ammonia synthesis activities of the tested materials and representative catalysts with physicochemical information, including metal particle sizes and surface areas. The promoted-Fe catalyst, which was first found by Mittasch more than 100 years ago and has been improved since then, is not inferior to most of the recently reported catalysts; only a handful of recent catalysts surpass the iron catalyst in terms of ammonia synthesis activity (Supplementary Table S1).8,16-18 Promoted-Fe, for example, has 7 times higher activity than Ru/C12A7:e– reported as a highly active catalyst based on a new approach (Supplementary Table S1). Nevertheless, promoted-Fe could not form ammonia below 200 °C (Fig. 2a). Detectable ammonia formation from the catalyst was barely observed above 220 °C. Cs-Ru/MgO and Ru/C12A7:e– also did not work for ammonia synthesis below 200 °C. In contrast, BaH2-BaO/Fe first revealed that iron can catalyze ammonia synthesis at 100 °C. The rate of ammonia formation increased with the reaction temperature. Ammonia formation proceeded over BaH2-BaO/Fe without a significant decrease in activity for over 100 h (Fig. 2b). Taking into account the number of zero-valent Fe atoms on the iron surface of the catalyst, the reaction for 100 h was estimated to give a turnover number of ca. 600 thousand (6.12×105); BaH2-BaO/Fe with a new type of catalyst structure acts as a stable catalyst.
Another notable feature in Fig. 2a is the catalytic activity and efficiency for ammonia formation of Ru/BaH2-BaO. Ru/BaH2-BaO above 300 °C is of the highest standard, as shown in Supplementary Table S116. However, Ru/BaH2-BaO could not synthesize ammonia at 100 ℃ and the ammonia formation rate was inferior to BaH2-BaO/Fe below 200 °C, even although both catalysts use the same electron-donating material. The difference is further emphasized with respect to the ammonia formation turnover for the surface zero-valent Fe and Ru atoms (turnover frequency; TOF). In Fig. 2a, BaH2-BaO/Fe exhibits TOFs that are several thousand times higher than Ru/BaH2-BaO. In heterogeneous catalysts that adopt Ru, Co and Ni as the active sites for ammonia synthesis, the maximum TOF is at most less than 0.17 s-1, even at 400 °C (Supplementary Table S2),11 which is too low compared with those of nitrogenase (0.7‒2.0 s-1)19 used for biological ammonia synthesis, although we cannot simply compare both ammonia syntheses. However, the TOF of the iron catalyst was 0.23 s-1 at 100 °C and reached 12.3 s-1 at 300 °C (Supplementary Table S2). These results clearly indicate that the iron catalyst is distinct from other transition metal-based catalysts. To understand the difference in catalysis between Fe and Ru promoted by the same electron-donating material, the correlation of ammonia formation rate with total pressure and the reaction orders in the rate equation on BaH2-BaO/Fe and Ru/BaH2-BaO were measured at 200 °C and are summarized in Fig. 3a. The ammonia formation rate for Ru/BaH2-BaO is independent of the pressure or gradually decreased with the pressure, and the reaction order for H2 showed a negative value of -0.79; an increase in reactant concentration cannot lead to an increase in product formation. The results are inconsistent with kinetics theory and are clearly due to hydrogen-poisoning13,14, where the Ru surface on Ru/BaH2-BaO is severely poisoned by H adatoms and cannot exhibit satisfactory catalytic performance for ammonia formation as shown in Supplementary Fig. S1c. Such a hydrogen-poisoned Ru surface would not show a high TOF. On the other hand, the rate of ammonia formation increased in proportion to the pressure over BaH2-BaO/Fe to show a positive value of +0.47 as the reaction order for H2. As a result, the iron catalyst is not strongly affected by hydrogen poisoning, even at 200 °C, which results in a much higher TOF than those of Ru-based catalysts well-known to be subject to hydrogen-poisoning.
One possible explanation for hydrogen-poisoning due to the H2 adsorption/desorption equilibrium on transition metals is the density and strength of the bonds between H adatoms and surface transition metal atoms. Because a larger amount of H adatoms is expected to be more tightly adsorbed on transition metal surfaces with a decrease in temperature, we may clarify the effect of hydrogen-poisoning by observing H2 desorption from BaH2-BaO/Fe and Ru/BaH2-BaO at low temperature. Figure 3b shows molecular deuterium (D2)-TPD measurements for both the catalysts. The catalysts were heated at 300 °C in a flow of Ar to remove H adatoms from the Fe and Ru surfaces. After they were cooled down to 25 °C, D adatoms were adsorbed onto the Fe and Ru surfaces in a flow of D2 (15 mL min-1, 30 min) at the temperature. The desorption of D adatoms due to gas phase D2 can be distinguished from the desorption of H adatoms originated from H‒ anions in BaH2-BaO through transition metals by the use of D2-TPD. Most D adatoms desorbed as D2 from the iron catalyst at 40‒100 °C. It was confirmed that D2 desorption from the iron catalyst is consistent with H2 desorption from a single crystal iron surface15, which means that the iron surface can provide sufficient adsorption sites for N2 and N adatoms without a strong influence of hydrogen-poisoning under ammonia synthesis conditions above 100 °C. In the case of Ru/BaH2-BaO, a large D2 desorption peak was observed at 125‒225 °C; Ru binds to H adatoms more tightly than iron, and these H adatoms can cause hydrogen-poisoning on Ru over a broad temperature range. Co and Ni surfaces have also been found to strongly adsorb H adatoms that are desorbed as H2 at > 150‒200 °C as with Ru20-22. This implies that many transition metals used to synthesize ammonia can be significantly affected by hydrogen-poisoning at low temperatures and iron is an exceptional transition metal that prevents hydrogen-poisoning. The difference in hydrogen-poisoning may be expressed as the difference in TOF among BaH2-BaO/Fe and the other transition metal-based catalysts.
Reaction mechanism of ammonia formation over BaH2-BaO/Fe
Ammonia formation from H2 and N2 over heterogeneous catalysts is roughly divided into N2 cleavage into N adatoms and N-Hn species formation, and in most catalysts, the former step has been regarded as the rate-determining step23. In the present work, ammonia synthesis over BaH2-BaO/Fe was studied from the point of view of the energy diagram (Fig. 4a), which suggested that the rate-determining step is not the former step, but the latter step over BaH2-BaO/Fe, because of the facile cleavage of N2. Arrhenius plots for ammonia synthesis over the tested catalysts are shown in Supplementary Fig. S5, and the apparent activation energies calculated from the gradients of the plots that correspond to the activation energy in the rate-determining step of the overall chemical process are also summarized in Supplementary Table S2 with those of other catalytic systems. The apparent activation energy for ammonia synthesis over BaH2-BaO/Fe was only 40±5 kJ mol-1. It is not only much smaller than that of Cs-Ru/MgO (124 kJ mol-1) as a benchmark catalyst, but is also below that of Ba-Ru/Ca(NH2)2 (59 kJ mol-1) reported as a catalyst with the highest catalytic activity for ammonia synthesis (Supplementary Tables S1 and S2)8. The energy diagram for ammonia formation on the iron surfaces of BaH2-BaO/Fe is shown in Fig. 4a with the thermodynamic results such as the apparent activation energies for ammonia synthesis (Ea NH3) and the 14N2-15N2 isotropic exchange reaction. The Arrhenius plot for the 14N2-15N2 isotropic exchange reaction (14N2 + 15N2 ↔ 214N15N, Ea 14N2−15N2 = 86±5 kJ mol-1) is shown in Supplementary Fig. S6. It has been confirmed that there is no significant difference in apparent activation energy between N2 isotopic exchange reaction and N2 desorption (Ea 2N→N2) by the recombination of N adatoms on ammonia synthesis over heterogeneous catalysts24-26. For this reason, the rate-determining step of the N2 isotopic exchange reaction is the same as that for the recombination of N adatoms (Ea 14N2−15N2 = Ea 2N→N2). The dissociative adsorption heat of N2 into N adatoms (∆H N2→2N) on a pure iron surface has been estimated to be -70 to -100 kJ mol-1 27. Adsorbed N atoms have a positive electron affinity; electron transfer from BaH2-BaO to N adatoms through iron stabilizes these atoms, which results in a larger dissociative adsorption heat on BaH2-BaO/Fe than that on pure Fe24,28. The dissociative adsorption heat of N2 on the iron surface of BaH2-BaO/Fe (∆H N2→2N) would therefore be less than -70 to -100 kJ mol-1 (|∆H N2→2N| ≥ 70‒100 kJ mol-1). From these results, the activation energy for N2 cleavage (Ea N2→2N) on BaH2-BaO/Fe should be below 16±5 kJ mol-1 from Ea 2N→N2 and ∆H N2→2N (Ea N2→2N = Ea 2N→N2 + |∆H N2→2N|). This value is too small compared with the apparent activation energy (Ea NH3=40±5 kJ mol-1) for ammonia synthesis over BaH2-BaO/Fe if we regard Ea N2→2N as the apparent activation energy, i.e., the activation energy for the rate-determining step (Fig. 4a-A). Consequently, it is consistent that the rate-determining step of ammonia synthesis on BaH2-BaO/Fe is not N adatom generation from adsorbed N2, but the formation of subsequent N-Hn species such as NH, NH2 and NH3 (Fig. 4a-B).
However, N2 cleavage is not as easily achieved as the formation of N-Hn species in most catalysts. To decrease the activation energy for the former below those of the latter by weakening strong N≡N triple bonds requires electron donation from the electron-donating materials to the antibonding π* orbitals of adsorbed N2 via transition metal d-orbitals (i.e., back donation) to be boosted. The strong back donation would weaken the N≡N stretching vibration by elongation of the N≡N bond. In this study, Fourier transform infrared (FT-IR) spectroscopy measurements using N2 as a probe molecule were used to observe the N≡N stretching of N2 adsorbed on transition metal surfaces. In the FT-IR spectrum for N2 adsorbed on Ru nanoparticles deposited on Al2O3 (Ru/Al2O3) (Fig. 4b), the N≡N stretching band (νN2) appeared at 2150-2250 cm-1 (peak top: 2200 cm-1), which is much lower than that of gaseous N2 (2744 cm-1). The large red-shift indicates that the electron donation from Ru to the antibonding π* orbitals of adsorbed N2 elongates the N≡N bond. The νN2 band was further red-shifted to 2100-2200 cm-1 (peak top: 2175 cm-1) in the spectrum for N2 adsorbed on Ru/C12A7:e−, an exceptional catalyst to shift the rate-determining step from N2 cleavage to N-H species formation as with BaH2-BaO/Fe. The stronger electron donation from C12A7:e− to N2 molecules adsorbed on Ru facilitates the generation of N adatoms and reduces the activation energy for the former below those for the latter24. The νN2 bands generally appear at 2100-2300 cm-1 in highly active catalysts for ammonia synthesis7,29. Figure 4b also shows that νN2 for N2-adsorbed BaH2-BaO/Fe is observed as a broad band in the range of 2000-2175 cm-1, which is lower than of Ru/C12A7:e−, and disappears after the removal of N adatoms by evacuation. Thus, the iron catalyst is more electron-donating than Ru/C12A7:e− and can split adsorbed N2 molecules more easily than the latter; the activation energy for the formation of N adatoms from adsorbed N2 molecules is decreased below those of N-Hn species formation
As a summary of these results, the mechanism postulated for ammonia synthesis over BaH2-BaO/Fe is shown schematically in Fig. 5. Only BaH2-BaO/Fe, metallic iron particles loaded with a BaH2-BaO mixture, catalyzed ammonia synthesis at low temperatures in the tested catalyst designs that adopted iron as the active sites. The catalyst designs based on conventional supported metal catalysts, including the iron nanoparticles-deposited BaH2-BaO mixture (Fe/BaH2-BaO), were not effective for low temperature ammonia synthesis. Most H adatoms on the iron surfaces desorb as H2 below 100-150 °C (Fig. 3b), so that the H2 adsorption/desorption equilibrium is shifted toward H2 desorption at low temperatures (Fig. 5A). For this reason, the H adatom concentration is not so high over the entire reaction temperature range above 100-150 °C. The low H adatom concentration induces BaH2 in BaH2-BaO to exhibit strong electron-donating capability (Fig. 5, B and C). The iron surface in contact with BaH2-BaO can pull H atoms out from BaH2 and release them as H2 molecules (BaH2 ↔ Ba2+H–(2–x)e–x + x/2H2) (Fig. 5B). This forms barium hydride species with electrons and hydride defects (Ba2+H–(2–x)e–x), a strong electron-donating material (Fig. 5C). A high concentration of H adatom on the iron surface would prevent the generation of such barium hydride species and the subsequent ammonia formation. The shift of the H2 adsorption/desorption equilibrium toward H2 desorption on the iron surfaces does not only give the strong electron-donating capability to BaH2-BaO, but also provides sufficient adsorption sites for N2 molecules on iron surface (Fig. 5D). Electrons are transferred from the resulting barium hydride species (Ba2+H–(2–x)e–x) to N2 molecules adsorbed on iron surfaces. FT-IR measurements (Fig. 4b) revealed that the antibonding π* orbitals of adsorbed N2 are significantly stimulated by the strong electron donation. The strong electron donation facilitates the cleavage of adsorbed N2 to N adatoms (Fig. 5, D and E) and can decreases the activation energy for N2 cleavage than those of the subsequent steps for N-Hn species formation. As a result, the latter N-Hn species formation can be the rate-determining step in ammonia synthesis over the catalyst, unlike those of most ammonia synthesis catalysts. The generated N adatoms react with H adatoms to form ammonia through the formation of N-Hn species (Fig. 5F). These reactions on BaH2-BaO/Fe proceed at more than ca. 100 °C (Fig. 2a), while hydrogen-poisoning is prevented. On the other hand, other transition metals such as Ru have tightly adsorbed H adatoms on their surfaces that cannot be easily removed as H2 below 150‒200 °C. The H adatoms would reduce the number of the adsorption sites for N2 molecules and N adatoms, and affect the reaction over the entire temperature range.