Efficient CeO₂ and CeO₂–Al₂O₃ supports for Ru as 3rd generation ammonia synthesis catalysts: enhanced kinetic mechanism over commercial Ru/CeO₂

Abstract

Ceria (CeO₂) has been previously reported as a functional support for ruthenium (Ru) as an ammonia synthesis catalyst. However, lab-synthesized ceria materials usually present low surface areas, thereby limiting the generation of oxygen vacancies and the ammonia synthesis activity as a result of weak metal–support interactions. With the aim of overcoming this issue, we prepared, by a simple impregnation method, high surface area ceria and ceria–alumina supported Ru catalysts with improved ammonia synthesis performance at moderate temperatures. In this sense, lab-synthesized Ru/CeO₂ (with higher specific surface area and lower crystallinity than commercial ceria) showed stronger metal–support interactions than the commercial sample, which resulted in a superior global ammonia synthesis kinetic mechanism with more positive hydrogen reaction orders (i.e., more resistant to hydrogen inhibition) and significantly lower activation energies (46 vs. 61 kJ mol⁻¹). We found that the use of alumina as a structural support increased the surface area of ceria, thereby promoting the Ru–CeO₂ interaction and the catalytic performance. We analyzed the effect of the surface chemistry of two different commercial aluminas (acidic and basic) with similar surface areas. Basic alumina was found to increase the specific surface area of the catalyst to a larger extent as compared to acidic alumina. Thus, the Ru/CeO₂–Al₂O₃ catalyst with 50 wt% of basic alumina showed an ammonia synthesis activity of 1.9 mmol g⁻¹ h ⁻¹ at 400 °C and ambient pressure and an activation energy as low as 44.8 kJ mol⁻¹.

---

Introduction

The change in the global energy paradigm during the last few decades shows a scenario in which green ammonia (NH₃) could be used as a chemical platform for decentralized green hydrogen (H₂) plants.^1–3 The current industrial process producing NH₃ from H₂ is incompatible with the use of renewables since harsh temperature (400–600 °C) and pressure (20–60 MPa) conditions are required to produce ammonia.⁴ Such high temperatures are required to activate the N₂ molecule and break the stable N≡N triple bond. Thus, high pressures are on demand in order to circumvent both kinetic and thermodynamic limitations since the ammonia synthesis reaction is highly exothermic (ca. −92 kJ mol_NH3⁻¹).⁵

Therefore, to successfully complete the transition into the new global green hydrogen scenario, the ammonia synthesis process needs to overcome several challenges, mainly related to the catalytic reaction system.^6,7 In this context, several alternatives to the conventional thermocatalytic ammonia synthesis process are being proposed by the research community, including plasma,^8,9 mechanocatalytic synthesis,¹⁰ chemical looping,^11–13 photothermal catalysis¹⁴ and electrocatalytic synthesis.^15,16

However, given the early stage of research of these processes, the investigation efforts to design efficient catalysts for thermocatalytic ammonia synthesis under mild conditions have also grown exponentially within the last two decades.^17–19 The most recent studies are focused on the development of stable and efficient catalysts based on non-noble metals, although the complex synthesis methods and conditions required represent a difficulty for their application scalability.^20,21 In this sense, most novel catalysts mainly rely on the performance of complex and sometimes non-stable functional supports such as electrides, hydrides, nitrides or intermetallics,^{17–19,22–25} which limits their practical applicability.

Rare earth metal oxides have been reported to present excellent electronic properties under strong metal–support interaction (SMSI) conditions, i.e., they can serve as electron donors for the transition metal (TM), thereby promoting nitrogen dissociation, which is usually the rate determining step (RDS) of the process.²⁶ Cerium oxide, the most abundant rare earth oxide, forms oxygen vacancies upon reduction, thereby promoting the metal (e.g., Ru, Co) by forming new interfacial metal–CeO_2−x sites²⁷ for the ammonia synthesis and promoting metal electronic donation. Thus, Ru/CeO₂-based catalysts have been reported to be more efficient than 1st generation Fe-based and 2nd generation Ru-based catalysts such as Ru/MgO or Ru/C.¹⁹

However, commercial ceria usually suffers from low surface area, which reduces the extent of the metal–support interaction and ultimately results in apparent activation energies significantly higher as compared to catalysts based on novel supports such as electrides and hydrides.^17,18 Therefore, the design of ceria-based catalysts with higher surface area and optimum metal–support interaction can increase both the number and activity of Ru sites in the ammonia synthesis reaction. One of the methods to increase the surface of ceria is the use of a high surface area material as a structural support. In this sense, alumina (Al₂O₃) is a widely used structural support in many applications in the field of thermocatalysis, since it has optimal structural properties. With regard to the ammonia synthesis, Al₂O₃ has been used as a structural promoter for Fe and some attempts were also made for Ru. However, it was concluded that acid sites from alumina can strongly interact with the NH₃ molecule, hindering its desorption.²² Therefore, basic supports, such as rare earth-based materials, are usually preferred as structural promoters.²⁸ Also, considering that the SMSI between Ru and CeO₂ mainly takes place on the catalyst surface,²⁹ it could be interesting to substitute a fraction of the ceria-based support with cheaper alumina, as long as the support surface could be covered by ceria.

In this work, we analyse the differences in terms of performance between two different materials as supports for Ru: commercial ceria (to be named as CeO₂|_C) and cerium oxide made by a simple impregnation–calcination method from cerium nitrate (to be named as lab-synthesized cerium oxide: CeO₂|_AS). We demonstrate that the lab-synthesized ceria outperforms the commercial ceria as a result of its higher surface area and lower crystallinity, which led to the formation of a higher concentration of more reactive surface oxygen vacancies. The impregnation–calcination synthesis method used herein allowed preparation of a ceria support with improved kinetic properties in the ammonia synthesis reaction as compared to commercial ceria.

Furthermore, we study the impact of the structural promotion of ceria with two types of different Al₂O₃ materials using a similar impregnation–calcination procedure. We found that the acid sites of γ-Al₂O₃ played a detrimental role in the performance of the Ru/CeO₂–Al₂O₃ catalysts, whereas basic Al₂O₃ allowed the kinetic performance of Ru/CeO₂ to be improved.

Materials and methods

Catalyst preparation

The lab-synthesized ceria support was prepared using an aqueous solution of Ce(NO₃)₃·6H₂O (99 wt% trace metal basis, Sigma-Aldrich). After the complete dissolution of cerium nitrate, the solution was evaporated overnight and the resulting solid was calcined at 400 °C for 6 h (2 K min⁻¹). Cerium oxide was softened by hand-milling with an agate mortar.

Ceria–alumina supports were prepared by aqueous impregnation of Ce(NO₃)₃·6H₂O in an aqueous suspension of any of the two different aluminium oxide materials: basic alumina (activated basic, Brockmann I, Sigma-Aldrich) and acidic alumina (anhydrous γ-alumina, Merck). The impregnation was carried out by stirring the suspension at room temperature for 6 h. After the impregnation and subsequent overnight evaporation, the support materials were calcined at 400 °C for 6 h (2 K min⁻¹) and then softened by hand-milling with an agate mortar.

Ruthenium(III) acetylacetonate (Ruacac 97%, Sigma-Aldrich) was used as the Ru precursor. The catalysts with a 5 wt% loading of Ru were prepared by impregnation of Ruacac onto the supports previously synthesized in ethanol. After impregnation, the suspensions were evaporated, and the catalysts were homogenized using an agate mortar. A pre-reduction treatment was applied to all the catalysts, following previously reported methods.³⁰ During the pre-reduction stage, the catalyst was treated under a pure H₂ flow at 400 °C for 3 h, following a heating rate of 10 °C min⁻¹. Then, a treatment with a mixture of N₂ + H₂ = 1 : 3 was carried out for 15 h.

Kinetic experiments

The kinetic tests were carried out in a micro-reaction system with automated integral pressure, temperature, gas flow and composition control (Microactivity-Effi, PID, Micromeritics) in a 316SS fixed-bed reactor. A mixture of N₂/H₂ = 1 : 3 with a total gas flow of 60 N mL min⁻¹ was set up, with a catalyst loading of 0.1 g and a space velocity of 36 000 mL g⁻¹ h⁻¹. To ensure that the reaction conditions are far away from equilibrium limitations, the base case is set up at 400 °C and atmospheric pressure. For the calculation of N₂ and H₂ reaction orders, He was used as an inert diluent gas. With the aim of obtaining accurate temperature measurements, a thermocouple was placed into the catalyst bed. The produced ammonia was trapped in an aqueous solution of diluted sulfuric acid (5 mM) and the ammonium ion (NH₄⁺) concentration was measured by means of an ion chromatograph (Dionex Easion, Thermo Scientific).

The reproducibility of the kinetic experiments was checked as follows: a repetition of the whole kinetic mapping for each catalyst was carried out in different runs until the values of the ammonia reaction rate under the base case conditions and the activation energies showed a standard deviation below 10% over the average.

The procedure for the calculation of the reaction orders is shown in the ESI,† section S1.

Characterization

Nitrogen physisorption experiment measurements (ASAP 2020 Plus, Micromeritics) were used to obtain the Brunauer–Emmett–Teller (BET) specific surface areas of the catalysts. Crystalline patterns were analysed by means of X-ray diffraction (XRD, Bruker D8-Advance) with Cu Kα radiation. X-ray fluorescence spectroscopy (XRF, Zetium de PANalytical) was carried out to measure the bulk composition of all the catalysts. Surface analyses were done by means of X-ray photoelectron spectroscopy (XPS, NEXSA, Thermo-Scientific), using monochromatized Mg Kα radiation. Binding energies were calibrated with the carbon C–C 1s peak (284.6 eV). H₂ temperature programmed reduction (H₂-TPR) analyses were done to study the SMSI and the formation of oxygen vacancies. A gas flow of 100 mL min⁻¹ with a composition of 5% H₂ into He was used, with a heating ramp of 10 °C min⁻¹. The outlet gas composition was measured using a mass spectrometer (Omnistar GSD 301 O₂, Pfeiffer Vacuum). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) experiments were conducted using a Talos F200X (Thermo Fisher Scientific) microscope, with high-resolution scanning (HRSTEM: 0.16 nm @200 kV). CO₂-TPD was carried out on a thermogravimetric analyzer (TGA SDT650, TA Instruments) coupled with a mass spectrometer (ThermoStar, Pfeiffer Vacuum). 35–60 mg of catalysts were loaded in a 90 μm alumina crucible. All the catalysts were pre-treated under pure Ar to 500 °C, after which the sample was cooled down to room temperature. Once at room temperature, CO₂ was fed to the reactor for 2 h to allow adsorption. Then, the temperature was raised to 40 °C under pure Ar to allow removal of the physisorbed CO₂. Finally, the TPD was conducted from 40 to 800 °C with a heating rate of 10 °C min⁻¹.

Inductively coupled plasma atomic emission spectroscopy (ICP-OES) experiments were conducted using the system Optima 7300 DV with dual vision (Perkin Elmer) in order to measure the Ru content of the catalysts and compare with the XRF results. Prior to the ICP experiments, the samples were dissolved in a mix of nitric acid and hydrochloric acid (1 : 3) under microwave conditions. Then, a scaling temperature program was followed up to 250 °C. After cooling down, the samples were filtered and diluted in ultrapure water.

Results and discussion

Lab-synthesized ceria vs. commercial ceria

The N₂ physisorption results for both commercial and lab-synthesized ceria-based catalysts are shown in Fig. 1a. The simple impregnation–calcination synthesis method used herein resulted in a material with a significantly higher specific surface area as compared to the commercial sample (76.3 vs. 34.0 m² g⁻¹). This difference in the specific surface area is important since the catalytic activity of Ru-based catalysts in this reaction has been found to be strongly influenced by the metal shape and particle size,^31,32 as well as by the support morphology.²⁹ Interestingly, the specific surface area obtained for Ru/CeO₂|_AS was higher than other typical ceria-based catalysts (Table S1†), which highlights the success of our simple impregnation–calcination synthesis.

Fig. 1 (a) Specific surface area of Ru/CeO₂|_C and Ru/CeO₂|_AS. (b) Ammonia synthesis rates at 0.1–0.9 MPa and 400 °C. (c) Ammonia synthesis rates at 340–400 °C and 0.1 MPa. (d) Nitrogen reaction order (α) plot, determined at 400 °C and 0.1 MPa. (e) Hydrogen reaction order (β) plot, determined at 400 °C and 0.1 MPa. (f) Ammonia reaction order (γ) plot, determined at 400 °C and 0.1 MPa.

The ammonia synthesis rates for both catalysts at low and high pressure are shown in Fig. 1b. Ru/CeO₂|_AS showed slightly higher activity than Ru/CeO₂|_C at 0.1 MPa (2.43 vs. 2.20 mmol g⁻¹ h⁻¹). Both values were higher than those reported for other Ru/CeO₂ catalysts under the same operating conditions.²⁷ The performance gap between both catalysts further increased to 20% at higher pressures (5.16 mmol g⁻¹ h⁻¹ for Ru/CeO₂|_AS and 4.31 mmol g⁻¹ h⁻¹ for Ru/CeO₂|_C). This pressure effect is intimately related to the kinetic efficiency, in particular, to the resistance towards hydrogen poisoning.³³ Thus, these results revealed a more efficient global kinetic mechanism for Ru/CeO₂|_AS.^21,34

The effect of the operating temperature on the ammonia synthesis performance was studied by performing runs at 400 (base case) and 340 °C (Fig. 1c). The relative difference in terms of performance between both catalysts was 4-fold higher at 340 °C, as compared to that observed at 400 °C. The Arrhenius analysis revealed an activation energy of 61.3 kJ mol⁻¹ for Ru/CeO₂|_C (Fig. S4b†), while a noticeably low value of 46.1 kJ mol⁻¹ was obtained for Ru/CeO₂|_AS (Fig. S4f†). This activation energy was significantly lower than those reported for other similar Ru/CeO₂ catalysts (Table S1†) and comparable to those of the best ammonia synthesis 3rd generation catalysts such as complex single atom catalysts (SACs), intermetallics and hydrides.^17–19,27 This outstanding performance highlights the success of the simple synthesis method used herein.

In the case of those catalysts following a dissociative mechanism, such as Ru/CeO₂, it is well known that the lower the activation energy, the more efficient the N₂ dissociation and activation step, which is usually the RDS of the reaction.^18,35–37 This was further confirmed by the N₂ reaction order analysis. Typical N₂ (α), H₂ (β) and NH₃ (γ) reaction order plots are shown in Fig. 1d–f, respectively. While β and γ were similar for both catalysts, α was 29% lower for Ru/CeO₂|_AS (ref. 35) compared to Ru/CeO₂|_C, revealing a lower importance of the N₂ dissociation step for Ru/CeO₂|_AS. It is also remarkable that β was positive for both catalysts, which suggests that hydrogen poisoning is not relevant over these catalysts, in line with the positive pressure effect shown in Fig. 1b.³⁸ Interestingly, Ru/CeO₂|_AS showed higher β values than the commercial sample (0.62 vs. 0.55, Fig. 1e), thereby revealing a higher resistance to hydrogen inhibition over these catalysts. Moreover, the highly negative values of γ (ca. −0.8) indicate that NH₃ is prone to be adsorbed over the catalysts, so the formation of NH_x species (rather than the N₂ dissociation step) plays a crucial role in the global kinetic scheme.³⁸

With the aim of gaining insight into the reasons for the different performances of Ru/CeO₂|_AS and Ru/CeO₂|_C, further characterization was carried out. Since the specific surface area of Ru/CeO₂|_AS was significantly higher than that of the commercial sample, it is expected that ceria has a lower particle size in this sample, leading to an improved metal–support interaction upon reduction.³⁹

An analysis of the Ru dispersion was carried out by means of HAADF-STEM with EDX mapping (Fig. 2a and b). Ru was found to be well dispersed over CeO₂ for both catalysts, although the lab-prepared sample showed optimum dispersion results (Fig. 2b). The HAADF-STEM pictures of all the catalysts characterized in this work are shown in Fig. S2,† whereas the EDX mapping with the contrast between Ru and Ce is shown in Fig. S3.† Ru/CeO₂|_C (Fig. S2a†) was found to present an organized structure in the form of nanoplatelets, whereas Ru/CeO₂|_AS (Fig. S2e†) seemed to be more amorphous. The Ru/CeO₂|_AS sample showed minor Ru agglomeration, which can be ascribed to the simplicity of the synthesis method.

Fig. 2 Ru coupled to Ce elemental EDX mapping from HAADF-STEM results of the (a) commercial ceria catalyst and (b) lab-synthesized ceria catalyst. (c) XRD patterns. (d) H₂-TPR profiles. (e) Ru 3p XPS spectra.

These results were further confirmed by XRD (Fig. 2c). The crystalline pattern showed peaks corresponding to the fluorite structure of CeO₂, which means that the support preserves the original Ce⁴⁺ crystalline lattice after the pre-reduction process. Ru/CeO₂|_AS showed less intense and wider diffraction peaks than Ru/CeO₂|_C, revealing lower crystallinity. In order to quantify the difference in the crystallinity between both catalysts, the crystallite sizes of the first peak i.e., 2θ = 28.5°, were obtained using the Scherrer equation shown in eqn (1).

where “D_C” is the crystallite size, “K” is the shape factor (a value of 0.9 was taken as an approximation in this case), “λ” is the X-ray wavelength, “β” is the full width at half maximum (FWHM) of the peak and “θ” is the peak angle (half of 2θ). Since the key point is a comparison between the crystallite sizes for both catalysts, an arbitrary value of 0 was taken for the instrumental line broadening, which leads to an approximation of the values of the crystallite sizes. As expected, RuCeO₂|_C showed a significantly higher CeO₂ crystallite size than the RuCeO₂|_AS sample (22.8 vs. 9.4 nm). As reported in previous studies, the lower the crystallinity, the higher the presence of structural defects. These results are in line with the improved kinetic mechanism for ammonia synthesis of the lab-synthesized sample as a result of an optimum Ru–CeO₂ contact.⁴⁰

The surface oxygen vacancies generated upon hydrogen reduction at 400 °C promoted the formation of interfacial metal–CeO_2−x active sites²⁷via metal–support interactions. In order to gain insight into the Ru–CeO₂ interaction, H₂-TPR experiments were conducted, from which the generation of oxygen vacancies and the specific role of the interactions between Ru and CeO₂ can be inferred. The H₂ (m/z = 2) profiles for both catalysts are shown in Fig. 2d. No reduction peaks were found at temperatures typical of the reduction of Ru oxides⁴¹ (below 180 °C), revealing that Ru is in the form of Ru⁰. The fact that Ru is not present as oxidized species can explain the absence of large Ru particles and agglomerates revealed by TEM (Fig. 2a and b). Thus, the preparation method used herein led to well dispersed Ru particles avoiding the formation of Ru oxide patches which are typical of impregnation–calcination processes.²² Two main reduction peaks were observed in the TPR profiles: one peak at low temperature (<250 °C) was attributed to the reduction of surface ceria in intimate contact with Ru, whereas the peak at higher temperature (>250 °C) corresponded to the reduction of surface ceria not in close contact with Ru. The reduction of surface ceria in close contact with Ru can be associated with SMSI between Ru and CeO₂.²⁹ By comparing the TPR profiles of the catalysts and the supports (Fig. S4a†), it can be inferred that Ru improved the reducibility of ceria, most likely by a hydrogen spillover process.⁴² Ru/CeO₂|_AS showed lower reduction temperatures for surface CeO₂ in contact with Ru as compared to Ru/CeO₂|_C (191 vs. 217 °C, Fig. 2d). These results revealed a more intense metal–support interaction for Ru/CeO₂|_AS and, potentially, an enhanced electron transfer between CeO₂ and Ru, which could account for its superior activity and the different kinetic mechanisms of both catalysts.⁴⁰ Reduction of the surface CeO₂ not in close in contact with Ru took place at a similar temperature for both catalysts (257 °C).

X-ray photoelectron spectroscopy (XPS) analyses were conducted to analyse surface Ru species (Fig. 2e). The Ru XPS spectra revealed the presence of two bands, namely, 3p_1/2 attributed to Ru³⁺ species⁴³ and 3p_2/3 assigned to Ru⁰.⁴⁴ Thus, XPS results revealed the presence of oxidized Ru species which probably formed upon air exposure prior to XPS analyses. We note that, prior to the reaction kinetic measurements, the catalysts were pre-reduced in situ at 400 °C for 3 h to ensure that no oxidized Ru species were present on the catalyst surface before the reaction. This in situ pre-reduction step was also carried out before the TPR experiments such that no reduction peaks originating from Ru species are expected to appear in the TPR profiles.

In conclusion, the simple impregnation–calcination method used herein led to the formation of a low crystallinity cerium oxide with a higher surface area, as compared to the commercial ceria. Ru/CeO₂|_AS showed a high degree of crystalline defects, which promote the formation of more active surface oxygen vacancies for those CeO₂ active sites in contact with Ru, showing an enhanced SMSI. This enhanced catalyst showed an outstanding performance for low-temperature ammonia synthesis (e.g., low activation energy) and a more efficient kinetic mechanism.

Structural promotion with basic and acidic alumina

As explained in the previous section, the formation of a high surface area low-crystallinity cerium oxide resulted in a Ru/CeO₂ catalyst with improved metal–support interaction and excellent low-temperature ammonia synthesis performance. With the aim of increasing the surface area of the cerium oxide and enhancing the Ru–CeO₂ interaction, we used alumina as a structural support. Two different aluminium oxide materials with similar specific surface areas (basic and acidic) were used as structural promoters.

The nomenclature of the samples is shown in Table 1, with samples S1 and S5 being respectively Ru/CeO₂|_C and Ru/CeO₂|_AS described in the previous section. The specific surface area of the samples is shown in Fig. 3a as a function of the ceria loading. As shown in Fig. 3a, the utilization of small amounts of alumina (e.g., 20 wt%) resulted in a significant increase of the BET surface area, especially for basic alumina. In the case of acidic alumina, the specific surface area of the catalyst remained nearly unchanged as the alumina loading increased from 20 to 80 wt%. The N₂ isotherms for the specific surface area determination are shown in Fig. S1.†

Table 1 Nomenclature of Ru/CeO₂–Al₂O₃ catalysts

Sample	Alumina	Ceria	CeO2 (wt%)
S1	—	Commercial	100
S2	Acidic	Lab-synthesized	20
S3			50
S4			80
S5	—		100
S6	Basic		20
S7			50
S8			80

---

Fig. 3 (a) Specific surface area of Ru/CeO₂–Al₂O₃ catalysts as a function of the ceria loading. (b) Ammonia synthesis rates in the base case conditions as a function of the ceria loading. (c) Activation energies at 400 °C and 0.1 MPa as a function of the ceria loading. (d) Nitrogen reaction order at 400 °C and 0.1 MPa as a function of the ceria loading. (e) Hydrogen reaction order at 400 °C and 0.1 MPa as a function of the ceria loading. (f) Long-term performance at 400 °C and 0.1 MPa.

The activity of the Ru/CeO₂–Al₂O₃ catalysts in the ammonia synthesis reaction was measured as a function of the ceria loading (Fig. 3b) under base case conditions (400 °C and 0.1 MPa). In the case of basic alumina, the ammonia synthesis rate increased continuously with the ceria loading, with a maximum value of 2.18 mmol g⁻¹ h⁻¹ for the sample with 80 wt% of ceria. In the case of the acidic alumina, the sample with 50 wt% of ceria showed an optimum activity (2.28 mmol g⁻¹ h⁻¹) and only slightly lower activity than the pure ceria sample Ru/CeO₂|_AS (S5) despite containing half of the ceria loading. These results also revealed that ceria plays a key role in promoting the ammonia synthesis reaction on Ru. In fact, once the ceria loading was reduced to 20 wt%, the ammonia synthesis rate decreased considerably, reaching values in the 0.5–0.7 mmol g⁻¹ h⁻¹ range for both basic and acidic alumina catalysts. This decrease can be explained by Ru–Ce sites being partially replaced with Ru–Al sites, which are not active considering that alumina is not a functional support for Ru.⁴⁵

Significant differences were found in the kinetic behaviour of the catalysts for both aluminas. Thus, the catalysts supported on acidic alumina showed higher activation energies than those supported on basic alumina and Ru/CeO₂|_AS (Fig. 3c). The Arrhenius plots for all the catalysts are shown in the ESI† (Fig. S3b–i). Interestingly, the activation energy of the 20 wt% ceria catalyst supported on acidic alumina was as high as 73.0 kJ mol⁻¹ (versus 53.7 kJ mol⁻¹ for the same catalyst supported on basic alumina). This result seems to indicate that the acid sites of γ-alumina have a detrimental effect on the Ru–CeO₂ interaction, thereby increasing the apparent activation energy for the reaction. Also, as previously reported, acid sites on the support strongly adsorb the ammonia produced (i.e., ammonia poisoning), negatively affecting the global catalyst performance.^28,46 Basic alumina catalysts slightly outperformed Ru/CeO₂|_AS in terms of an enhanced kinetic mechanism, with an optimum activation energy of 44.8 kJ mol⁻¹ for a ceria loading of 50 wt%. This result is interesting in that 50% of ceria can be replaced with cheaper alumina with a minimum effect on the activity and an improved kinetic mechanism.

Nitrogen (α), hydrogen (β), and ammonia (γ) reaction orders are shown in Fig. S5a–c,† respectively. The dependence of α and β on the ceria loading is shown in Fig. 3d and e, respectively. For both acidic and basic alumina, the nitrogen reaction orders reached minimum values at a ceria loading of 50 wt% (0.69 and 0.60, respectively). Overall, catalysts supported on basic alumina showed lower nitrogen reaction orders than catalysts supported on acidic alumina. These results are in line with the Arrhenius results (Fig. 3c) revealing a facilitated nitrogen dissociation and activation pathway for catalysts supported on basic alumina. Noticeably, the 50 wt% ceria catalyst supported on basic alumina showed lower nitrogen reaction orders than Ru/CeO₂|_AS (0.60 vs. 0.74). With regard to hydrogen reaction orders, positive values were obtained for ceria loadings higher than 50 wt%, thereby ruling out hydrogen inhibition (unlike catalysts with 20 wt% of ceria).

All the catalysts tested in this work showed excellent long-term stability (Fig. 3f). After an initial activation stage of ca. 24 h, variations in the ammonia synthesis rate below 10% under base case conditions (400 °C and 0.1 MPa) were found in all cases for more than 100 h on stream.

The HAADF-STEM pictures of the alumina-supported catalysts are shown in Fig. S2.† Unlike Ru/CeO₂|_C, none of the Ru/CeO₂–Al₂O₃ catalysts seem to present a clearly defined structure, probably because of the poor crystallinity of both aluminas after the calcination process (Fig. S5d†). Elemental EDX patterns of Ru, Ce and La (Fig. S3†) revealed poorer Ru dispersion for the alumina-based catalysts as compared to Ru/CeO₂|_AS (Fig. S3e†).

The XRD patterns of the CeO₂–Al₂O₃ samples (Fig. 4a) showed no evidence of Al₂O₃ phases, in line with the poor crystallinity of these materials. CeO₂ crystallite sizes (calculated by the Scherrer equation, Fig. 4b) similar to those of Ru/CeO₂|_AS were found for ceria loadings of 50 and 80 wt%, with samples supported on basic alumina showing slightly lower ceria crystallite sizes than their acidic alumina-supported counterparts. However, this difference seems to be marginal to explain the differences in terms of performance of both aluminas. Overall, the CeO₂ crystallite size for all the ceria–alumina based catalysts does not account for the different kinetic behaviour explained above. A simulation of the reference patterns from CeO₂, Al₂O₃ and Ru found in the database is shown in Fig. S5e† for clarity.

Fig. 4 Characterization of Ru/CeO₂–Al₂O₃ catalysts: (a) XRD patterns and (b) CeO₂ crystallite sizes. (c) Surface ceria content as a function of ceria loading determined by XPS. (d) Surface Ce³⁺ concentration as a function of ceria loading determined by XPS. (e) H₂-TPR profiles. (f) CO₂-TPD profiles. (g) Proportion of metal lattice oxygen species O_l from the O 1s XPS spectra as a function of the concentration of Ce on the catalyst surface.

As aforementioned, the SMSI is related to the Ru–Ce interactions at the catalyst surface. In order to study the distribution of the different elements on the catalyst surface and their oxidation state, XPS analyses were conducted. The distribution of Al, Ce and Ru was calculated assuming that these elements were present on the surface in the form of Al₂O₃, CeO₂ and Ru⁰, respectively, as an approximation, since those are the expected chemical species for Al, Ce and Ru, respectively. The amount of CeO₂ on the surface as a function of the ceria loading is shown in Fig. 4c. For acidic alumina catalysts, there is a direct relationship between the ceria loading and the amount of ceria present on the catalyst surface, with maximum deviations of 12 wt% between both parameters, thus revealing that cerium oxide is well dispersed over the alumina matrix. Conversely, the basic alumina catalysts with 50 wt% loading showed the highest deviation between the ceria contents on the catalyst surface and in the bulk, which could explain the good kinetic properties of this catalyst (e.g., low activation energy, low nitrogen reaction order and positive hydrogen reaction order). In order to gain insight into the reducibility of the ceria on these catalysts, a previously reported method was used to calculate the distribution of Ce³⁺ and Ce⁴⁺ from the deconvoluted XPS patterns.⁴⁷ These results are shown in Fig. 4d, whereas the Ce 3d XPS patterns are shown in Fig. S6a.† For basic alumina catalysts, the lower the ceria loading, the higher the amount of surface Ce³⁺. In the case of acidic alumina catalysts, the amount of surface Ce³⁺ reached a maximum for a ceria loading of 50 wt%, in line with the kinetic results revealing an optimum performance for this catalyst among the acidic alumina-based materials. The XPS patterns for Ru are shown in Fig. S6b.†

We have analysed the XPS results of our catalysts. In particular, the surface oxygen species on the catalysts were studied by analysing the O 1s spectra. Following the methodology of previous reports,^48,49 we have deconvoluted the O 1s band into three main peaks (Fig. S7†), which can be ascribed to different oxygen species. Thus, the peak at low binding energy (ca. 529.1–530.8 eV, O_l) is usually attributed to ceria lattice oxygen species in close contact with Ru (Ru–O–Ce), the second peak at ca. 530.9–532.0 eV is ascribed to ceria surface oxygen vacancies (O_v), while the third peak (ca. 531.5–533.5 eV) can be ascribed to adsorbed oxygenated species such as water and oxygen (O_s). The first two species (O_l and O_v) are associated with the presence of Ru–CeO₂ interactions leading to both electronic and structural interactions that result in catalytic promotion of Ru in this reaction. We quantified the amount of these species from the deconvoluted XPS spectra and the results are reported in a new table (Table 2). Interestingly, as shown in Fig. 4g, the number of O_l species was found to be directly related to the surface concentration of CeO₂ determined by XPS. On the other hand, the amount of ceria oxygen vacancies (O_v) increased with the alumina content since the structural promotion of Al₂O₃ leads to higher CeO₂ dispersions on the catalysts (Table 2). Thus, there seems to be a trade-off between the number of Ru–O–Ce sites (which increases with the ceria loading) and the number of oxygen vacancies (which increases with the alumina loading). This trade-off could explain the optimum activation energy found for the 50% CeO₂ supported on basic alumina (Fig. 3c).

Table 2 Percentage amounts of O 1s species obtained by deconvolution of the XPS O 1s band

Sample	Ol (%)	Ov (%)	Os (%)
S1	86.0	13.8	0.2
S2	19.9	65.6	14.5
S3	57.8	37.1	5.1
S4	81.0	14.2	4.8
S5	88.3	2.5	9.1
S6	9.9	71.2	18.9
S7	70.3	21.4	8.3
S8	54.0	29.8	16.2

---

XRF (Fig. S6c†) revealed similar tendencies to those observed in the XPS results for the surface ceria content. Furthermore, a Ru effective loading of 2–3% was achieved for all the catalysts reported in this work, with no significant relationship between Ru and CeO₂ loadings (Fig. S6d†) for both acidic and basic alumina-based catalysts.

The Ru loading was also determined by ICP-OES and the results are reported in Table 3. ICP results slightly differed from those obtained by XRF (Fig. S4d†). With the exception of S1, ICP revealed Ru loadings in the range of 1–3 wt% for all the catalysts synthesized herein. We note that these values are lower than the nominal content (5 wt%), and this could be explained by the well-known recalcitrance of Ru towards acids, which often results in incomplete dissolution.^50,51

Table 3 Ru loadings determined by XRF and ICP-OES experiments

Sample	Nominal Ru loading (wt%)	XRF Ru loading (wt%)	ICP-OES Ru loading (wt%)
S1	5.0	2.2	4.3
S2		2.4	1.1
S3		2.1	2.3
S4		2.5	1.4
S5		2.5	1.6
S6		3.1	2.1
S7		2.7	2.4
S8		2.6	2.7

---

The H₂-TPR profiles of the ceria–alumina catalysts (Fig. 4e) showed two main peaks at low and high temperatures which can be ascribed to the reduction of surface ceria in close contact with Ru and surface ceria not in contact with Ru, respectively. The low temperature peaks appeared at lower temperatures as the ceria loading increased for both aluminas, with optimum values (207 and 219 °C) for 80 wt% ceria loading on acidic and basic alumina, respectively. Interestingly, alumina-supported catalysts showed reduction peaks at higher temperatures compared to Ru/CeO₂|_AS (S5, Fig. 4e), which seems to indicate that the presence of alumina affects the net formation of surface oxygen vacancies as a consequence of a less intense interaction between Ru and CeO₂, particularly for basic alumina. Therefore, it seems clear that the strength of the metal–support interaction for the CeO₂–Al₂O₃ catalysts depends highly on the ceria loading rather than on the surface area. For instance, the S8 sample showed reduction peaks at higher temperatures than Ru/CeO₂|_AS despite having significantly higher specific surface area.

We conducted CO₂-TPD experiments for all the catalysts tested in this work (Fig. 4f). The TPD profiles of the supports (basic alumina, acidic alumina, and lab-prepared CeO₂) are also reported in the ESI† (Fig. S6e). As shown in Fig. 4f, all the samples under study showed similar TPD profiles containing one single desorption peak at low temperatures (109–158 °C), thereby revealing weak surface basicity in all cases. This low-temperature desorption peak is normally ascribed to the decomposition of carbonates formed during CO₂ adsorption at room temperature, which takes place with the simultaneous release of CO₂ and H₂O (not shown). Both commercial (S1) and lab-synthesized (S5) Ru/CeO₂ samples showed surface sites with similar basicity strength (i.e., desorption peaks at the same temperature), although the amount of surface basic sites was significantly higher for the lab-synthesized sample, in line with its higher surface area (Fig. 1). Thus, the number of ceria surface sites (and thus the improved Ru–CeO₂ interaction) rather than the basicity strength seems to explain the different kinetic behaviour of both samples (Fig. 1). The TPD profiles of the alumina-containing samples with low CeO₂ loading (20 wt%, S2 and S6) showed profiles very similar to those of their alumina parent supports (Fig. S6e†). As expected, the S6 sample supported on basic alumina showed a CO₂ desorption peak at higher temperatures than the S2 sample supported on acidic alumina (158 vs. 129 °C). Finally, the samples with higher CeO₂ content (50–80 wt%, S3, S4, S7, and S8) showed TPD profiles very similar to that of high surface area Ru/CeO₂ (S5). In this case, the samples supported on basic alumina (S7 and S8) showed desorption peaks with higher area than their counterparts supported on acidic alumina (S3 and S4), thereby revealing a higher CeO₂ dispersion on basic alumina (in line with the CeO₂ crystallite size data of Fig. 4b). These results could also explain the superior kinetic behaviour of basic alumina-supported samples S7 and S8 in terms of activation energy as compared to their acidic alumina-supported counterparts (Fig. 3c).

As a summary, the role of alumina as a structural promoter for ceria in Ru/CeO₂–Al₂O₃ catalysts has been unveiled. The use of alumina as a support allowed a high surface area, low crystalline and highly dispersed reducible cerium oxide to be generated, which helped improve the ammonia synthesis mechanism by decreasing the activation energy and the nitrogen reaction order, with an optimum loading of 50 wt% for both acidic and basic alumina supports. Acidic alumina had a negative effect on the catalytic performance, probably due to the strong adsorption of the ammonia produced in acid sites (ammonia inhibition). This poisoning effect was most relevant for low ceria loadings (e.g., 20 wt%). The Ru/CeO₂–Al₂O₃ catalyst with 50 wt% ceria and basic alumina showed optimum kinetic performance with an activation energy as low as 44.8 kJ mol⁻¹.

Conclusions

A simple impregnation–calcination method was used herein for the synthesis of CeO₂ and CeO₂–Al₂O₃ supports for Ru. These catalysts showed an outstanding performance for the low-temperature ammonia synthesis reaction, with activation energies comparable to those of the best catalysts reported so far. In this sense, a lab-prepared high-surface-area sample Ru/CeO₂|_AS showed an activation energy as low as 46.1 kJ mol⁻¹. This sample also showed excellent low-temperature kinetic behaviour with a very positive H₂ reaction order (thereby avoiding typical hydrogen inhibition issues associated with Ru) and a significantly lower N₂ reaction order as compared to the commercial counterpart (thereby allowing enhanced N₂ dissociation/activation).

Structural promotion of ceria with alumina led to the formation of catalysts with higher specific surface areas and lower ceria crystallinity. While acid sites from γ-alumina were found to negatively affect the ammonia synthesis performance, basic alumina allowed the ceria loading to be reduced while maintaining excellent activity and kinetic behaviour (activation energy as low as 44.8 kJ mol⁻¹) and reaction orders comparable to those of Ru/CeO₂|_AS.

The good performance of these easily prepared catalysts paves the way for a new family of 3rd generation catalysts which can be used as an alternative to complex electrides, hydrides and intermetallics, thereby providing new solutions for easier industrial catalyst scale up and effective integration of ammonia synthesis technology with renewables. We note that several improvements could be made to our catalysts. First, electropositive alkaline promoters could be added to facilitate electronic transfer towards Ru, thereby facilitating N₂ dissociation by providing electrons to the antibonding molecular orbitals. Second, we could enhance the Ru–CeO₂ interaction by adding elements able to form a solid solution with CeO₂, such that the CeO₂ lattice is distorted and the formation of oxygen vacancies (e.g., CeO_x) is facilitated.

Data availability

The data supporting this article have been included as part of the ESI.†

Author contributions

Javier Arroyo-Caire: conceptualization, methodology, validation, formal analysis, investigation, writing – original draft, visualization. Edgar. S. Duran-Uribe: methodology, formal analysis, resources. Mayra Anabel Lara-Angulo: resources, writing – review & editing. Manuel Antonio Diaz-Perez: conceptualization, writing – review & editing, supervision. Antonio Sepúlveda-Escribano: writing – review & editing, supervision. Juan Carlos Serrano-Ruiz: conceptualization, formal analysis, investigation, writing – review & editing, supervision, project administration, funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This publication is part of the project BioEnH2, funded by MCIN/AEI/https://doi.org/10.13039/501100011033 and by European Union NextGenerationEU/PRTR. The authors want to acknowledge the University of Cádiz (UCA) for their services in the microscopy field and their gentle analyses and reports.

References

1. M. Reese, C. Marquart, M. Malmali, K. Wagner, E. Buchanan, A. McCormick and E. L. Cussler, Ind. Eng. Chem. Res., 2016, 55, 3742–3750.
2. A. Laval, Hafnia, H. Tøpsoe, Vestas and S. Gamesa, Ammonfuel. An industrial view of ammonia as a marine fuel, 2020.
3. K. H. R. Rouwenhorst, A. G. J. Van der Ham and L. Lefferts, Int. J. Hydrogen Energy, 2021, 46, 21566–21579.
4. C. Smith, A. K. Hill and L. Torrente-Murciano, Energy Environ. Sci., 2020, 13, 331–344.
5. O. A. Ojelade and S. F. Zaman, Chem. Pap., 2021, 75, 57–65.
6. Q. Wang, J. Guo and P. Chen, J. Energy Chem., 2019 DOI:10.1016/j.jechem.2019.01.027.
7. F. Chang, W. Gao, J. Guo and P. Chen, Adv. Mater., 2021 DOI:10.1002/adma.202005721.
8. M. L. Carreon, J. Phys. D: Appl. Phys., 2019, 52, 483001.
9. J. Hong, S. Prawer and A. B. Murphy, ACS Sustainable Chem. Eng., 2018, 6, 15–31.
10. S. Reichle, M. Felderhoff and F. Schüth, Angew. Chem., Int. Ed., 2021, 60, 26385–26389.
11. L. Li, T. Zhang, J. Cai, H. Cai, J. Ni, B. Lin, J. Lin, X. Wang, L. Zheng, C. T. Au and L. Jiang, J. Catal., 2020, 389, 218–228.
12. J. Moon, Y. Cheng, L. Daemen, E. Novak, A. J. Ramirez-Cuesta and Z. Wu, Top. Catal., 2021, 64, 685–692.
13. H. Yan, W. Gao, Q. Wang, Y. Guan, S. Feng, H. Wu, Q. Guo, H. Cao, J. Guo and P. Chen, J. Phys. Chem. C, 2021, 125, 6716–6722.
14. S. Wang, W. Yu, S. Xu, K. Han and F. Wang, ACS Sustainable Chem. Eng., 2022, 10, 115–123.
15. Y. Li, H. Wang, C. Priest, S. Li, P. Xu and G. Wu, Adv. Mater., 2021, 33, 2000381.
16. J. M. McEnaney, A. R. Singh, J. A. Schwalbe, J. Kibsgaard, J. C. Lin, M. Cargnello, T. F. Jaramillo and J. K. Nørskov, Energy Environ. Sci., 2017, 10, 1621–1630.
17. J. Humphreys, R. Lan and S. Tao, Adv. Energy Sustainability Res., 2021, 2, 2000043.
18. V. S. Marakatti and E. M. Gaigneaux, ChemCatChem, 2020 DOI:10.1002/cctc.202001141.
19. J. Arroyo-Caire, M. A. Diaz-Perez, M. A. Lara-Angulo and J. C. Serrano-Ruiz, Nanomaterials, 2023, 13, 2914.
20. C. Smith and L. Torrente-Murciano, Chem Catal., 2021 DOI:10.1016/j.checat.2021.09.015.
21. H. Hosono and M. Kitano, Chem. Rev., 2021 DOI:10.1021/acs.chemrev.0c01071.
22. N. Saadatjou, A. Jafari and S. Sahebdelfar, Chem. Eng. Commun., 2015 DOI:10.1080/00986445.2014.923995.
23. L. Li, T. Zhang, Y. Zhou, X. Wang, C. tong Au and L. Jiang, J. Rare Earths, 2022 DOI:10.1016/j.jre.2021.06.014.
24. Y. Nakaya and S. Furukawa, Chem. Rev., 2022, 123, 5859–5947.
25. J. Guo and P. Chen, Acc. Chem. Res., 2021, 54, 2434–2444.
26. K. Sato and K. Nagaoka, Chem. Lett., 2021, 50, 687–696.
27. Y. Gong, H. Li, C. Li, X. Bao, H. Hosono and J. Wang, J. Adv. Ceram., 2022, 11(10), 1499–1529.
28. S. ichiro Miyahara, K. Sato, Y. Kawano, K. Imamura, Y. Ogura, K. Tsujimaru and K. Nagaoka, Catal. Today, 2021, 376, 36–40.
29. Z. Ma, S. Zhao, X. Pei, X. Xiong and B. Hu, Catal. Sci. Technol., 2017, 7, 191–199.
30. M. Kitano, Y. Inoue, M. Sasase, K. Kishida, Y. Kobayashi, K. Nishiyama, T. Tada, S. Kawamura, T. Yokoyama, M. Hara and H. Hosono, Angew. Chem., 2018, 130, 2678–2682.
31. C. Fernández, C. Sassoye, D. P. Debecker, C. Sanchez and P. Ruiz, Appl. Catal., A, 2014, 474, 194–202.
32. Z. Song, T. Cai, J. C. Hanson, J. A. Rodriguez and J. Hrbek, J. Am. Chem. Soc., 2004, 126, 8576–8584.
33. P. Wang, F. Chang, W. Gao, J. Guo, G. Wu, T. He and P. Chen, Nat. Chem., 2017, 9, 64–70.
34. H. Hosono, Catal. Lett., 2022, 152, 307–314.
35. K. Ooya, J. Li, K. Fukui, S. Iimura, T. Nakao, K. Ogasawara, M. Sasase, H. Abe, Y. Niwa, M. Kitano and H. Hosono, Adv. Energy Mater., 2021, 11, 2003723.
36. M. Kitano, Y. Inoue, Y. Yamazaki, F. Hayashi, S. Kanbara, S. Matsuishi, T. Yokoyama, S. W. Kim, M. Hara and H. Hosono, Nat. Chem., 2012, 4, 934–940.
37. Y. Gong, H. Li, J. Wu, X. Song, X. Yang, X. Bao, X. Han, M. Kitano, J. Wang and H. Hosono, J. Am. Chem. Soc., 2022, 144, 8683–8692.
38. Y. Kobayashi, M. Kitano, S. Kawamura, T. Yokoyama and H. Hosono, Catal. Sci. Technol., 2017, 7, 47–50.
39. W. Li, P. Liu, R. Niu, J. Li and S. Wang, Solid State Sci., 2020, 99, 105983.
40. K. Sato, K. Imamura, Y. Kawano, S. ichiro Miyahara, T. Yamamoto, S. Matsumura and K. Nagaoka, Chem. Sci., 2016, 8, 674–679.
41. X. Wang, X. Peng, Y. Zhang, J. Ni, C. T. Au and L. Jiang, Inorg. Chem. Front., 2019, 6, 396–406.
42. Z. Feng, F. Guo, Y. Zhang, T. Ichikawa and J. Zheng, Appl. Catal., B, 2025, 125059.
43. B. Folkesson, M. Bjorøy, J. Pappas, S. Skaarup, R. Aaltonen and C.-G. Swahn, Acta Chem. Scand., 1973, 27, 287–302.
44. M. M. T. Khan and S. Srivastava, Polyhedron, 1988, 7, 1063–1065.
45. H. B. Kim and E. D. Park, Catal. Today, 2023, 411–412, 113817.
46. M. Osozawa, A. Hori, K. Fukai, T. Honma, K. Oshima and S. Satokawa, Int. J. Hydrogen Energy, 2022, 47, 2433–2441.
47. J. Silvestre-Albero, F. Rodríguez-Reinoso and A. Sepúlveda-Escribano, J. Catal., 2002, 210, 127–136.
48. Z. Peng, Y. Wang, C. Yin, S. Qiu, Y. Xia, Y. Zou, F. Xu, L. Sun and H. Chu, Sustainable Energy Fuels, 2023, 7, 821–831.
49. Y. juan Hao, Y. guang Ma, X. Zhang, J. Li, S. Wang, X. Chen and F. tang Li, Chem. Eng. J., 2022, 433, 134619.
50. M. Balcerzak, Crit. Rev. Anal. Chem., 2002, 32, 181–226.
51. T. Suoranta, M. Niemelä and P. Perämäki, Talanta, 2014, 119, 425–429.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5cy00122f

---