Dispersed surface Ru ensembles on MgO(111) for catalytic ammonia decomposition

The study addresses how to design efficient catalysts for ammonia decomposition to enable hydrogen storage and utilization with renewable energy. While Ru is known to be highly active, the exact nature of active sites and mechanisms remains unclear. Classical views emphasize Ru nanoparticles with B5-type step sites as optimal for N2 activation, but these suffer from hydrogen poisoning and poor atom economy. The authors hypothesize that atomically dispersed Ru on polar oxide supports, particularly MgO(111), can promote heterolytic N–H activation and hydrogen spillover, reducing poisoning and enabling N–N recombination via cooperative neighboring single-atom Ru sites, drawing analogies to multinuclear homogeneous N2 activation complexes.

Prior work shows Ru is active and structure-sensitive for ammonia decomposition, with B5 sites on Ru nanoparticles identified as key for N2 activation; their abundance depends on particle size and shape (maxima around 1.8–3 nm hemispherical and ~7 nm elongated particles). However, nanoparticle formation reduces atom economy and Ru catalysts often suffer hydrogen poisoning, with negative hydrogen reaction orders indicating site blockage. Supports that facilitate hydrogen spillover (e.g., C12A7:e−, barium hexaaluminate, CNTs) improve activity compared with inert supports (SiO2, Al2O3). Polar oxide surfaces can exhibit unique catalytic effects due to surface polarity and electrostatic properties. Recent advances have achieved atomic dispersion of transition metals on MgO. Homogeneous catalysis has demonstrated bi-/multinuclear metal complexes activating N2 via diazenido, hydrazido, or nitrido species, inspiring the exploration of cooperative surface single-atom ensembles. These insights motivate evaluating MgO facets and Ru loadings for maximizing activity via single-atom synergy and hydrogen spillover.

• Supports: MgO nanosheets exposing (111), (110), and (100) facets synthesized via hydrothermal (MgO(111) from MgCl2 with benzoic acid surfactant, NaOH precipitation, calcination at 500 °C), reconstruction of commercial MgO (boiling in water, drying, vacuum calcination) for (110), and thermal decomposition of Mg(NO3)2 for (100). MgO nanocubes (exclusive (100)) were combusted from Mg ribbon; controlled wet etching (pH 2–6.8) cleaved cubes to expose (110) and (111) facets, enabling series with varied facet ratios; vacuum calcination removed hydroxyls.
• Ru deposition: Ru3(CO)12 was dispersed in THF, contacted with MgO supports, solvent removed, precursors vacuum-treated at 90 °C, then vaporized/decomposed and deposited on MgO up to 300 °C under Ar to avoid MgO surface deconstruction; loadings varied 0.01–7 wt%.
• Catalytic testing: Fixed-bed flow reactor, 50 mg catalyst, pretreatment at 350 °C in 5% H2/He for 4 h. NH3 decomposition conducted at 200–600 °C, 1 bar, varied WHSV (typically 30,000 mL gcat−1 h−1; also up to 60,000). Gas analysis by GC-TCD (HayeSep Q). Kinetics: Arrhenius plots from rates at <15% conversion under identical flows; reaction orders by varying partial pressures of NH3, H2, N2 at 350 °C.
• Facet contribution deconvolution: Using 11 samples with known percentages of (100)/(110)/(111), the overall conversion was regressed as Conv = aX + bY + cZ to derive per-facet activities (contour maps at 300–400 °C).
• Characterization: XRD for phase/facet orientation; TEM/HAADF-STEM (including FFT and image simulations) to visualize Ru dispersion and measure Ru–Ru distances; XPS (Ru 3p/3d) for oxidation states; AP-XPS at 300 °C under 0.9 mbar NH3 or Ar to monitor dynamic changes in Ru, O (including OH), and N species; EXAFS at Ru K-edge for coordination environment (Ru–O, Ru–Ru); H2-TPD with MS; TPSR (NH3 adsorbed at 50 °C, Ar purge, ramp to 800 °C, monitor m/z 2, 17, 28); in situ DRIFTS of NH3 adsorption/decomposition (300–400 °C), time-resolved and cycling between NH3 and Ar to track OH and Ru–N/NH2 species.
• DFT: VASP (PBE-GGA, D3 dispersion), PAW potentials, 450 eV cutoff, 1×1×1 k-point for large slabs, forces <0.01 eV/Å. Slab models: MgO(111) (O-terminated top, Mg-terminated bottom), (100), (110) with 15 Å vacuum; bottom two layers fixed; dipole correction applied. Considered adsorbed Ru atop O3 hollow (Ru-OOO) as most stable versus embedded; Bader charge analyses; NEB for barriers. Models of two neighboring single Ru atoms (~5.04 Å apart) on each facet to study stepwise NH3 dehydrogenation and N–N recombination; separate single-Ru models for H spillover barriers. H2 desorption step excluded; effect of retained H* (on (100)/(110)) or spillover H–O (on (111)) considered in thermodynamics.

• Structure and oxidation state: On MgO(111), Ru is atomically dispersed, coordinated to three surface O atoms (Ru–O at 2.03 Å; Ru–Mg at 3.17 Å), with no Ru–Ru contribution in EXAFS. HAADF-STEM shows single Ru atoms and close-proximity 2× and 3× Ru surface ensembles with Ru–Ru separations ~5.1 Å (no metallic lattice). XPS Ru 3p3/2 at 463.3 eV indicates Ru2+ on MgO(111), while Ru/MgO(110) and (100) show lower binding energies (462.1, 461.9 eV) consistent with metallic Ru nanoparticles.
• Catalytic activity (400 °C, WHSV 30,000 mL gcat−1 h−1, 1 bar): Reaction rates (µmol NH3 gcat−1 s−1): Ru/MgO(111) 247.8; Ru/MgO(110) 119.8; Ru/MgO(100) 74.7; Ru/Com-MgO 92.3. TOF at 400 °C: 2.33 s−1 (111); 0.90 s−1 (110); 0.59 s−1 (100). At 3.1 wt% Ru on MgO(111), near-equilibrium conversion (98–99%) at 425 °C (WHSV 30,000). At 450 °C, WHSV 60,000, specific activity 1777.4 mmol H2 gRu−1 min−1 with TOF 4.91.
• Kinetics: Apparent activation energy Ea: 71.3 kJ mol−1 (111) vs 105.5 (110) and 119.2 (100). Reaction orders at 350 °C: hydrogen order γ is −0.47 (111) vs −0.79 (110) and −0.85 (100), indicating reduced H poisoning on (111); ammonia order α is larger for (111) (0.66), consistent with facilitated N–N recombination. N2 partial pressure has negligible effect, supporting N recombination as RDS.
• Transient spectroscopy: TPSR shows NH3 starts decomposing at ~198 °C on Ru/MgO(111) versus ~300 °C on Ru/MgO(100); higher H2 and N2 signals on (111). H2-TPD: strong high-T desorption (~475 °C) for (110)/(100) evidences strong H adsorption; (111) shows much lower intensity at high T with an additional shoulder near 200 °C, consistent with facile H transfer. In situ DRIFTS (111): immediate Ru–N band at ~2047 cm−1; growth of OH bands (stretch ~3300 cm−1, bend ~1900 cm−1), indicating proton spillover; (100) displays NH2 at 1573 cm−1 and lacks OH features. Cycling DRIFTS between NH3 and Ar shows reversible OH formation and removal, consistent with reversible H spillover and H2 formation.
• AP-XPS (300 °C, 0.9 mbar): Under NH3, Ru on MgO(111) reduces from Ru2+ to Ru0 (Ru 3d shifts 281.1→279.9 eV) while surface OH increases and N species appear; switching to Ar restores Ru2+ and diminishes OH and N signals, evidencing reversible proton hopping and N, H desorption.
• Facet contribution: Contour-map deconvolution across 11 samples shows (111) gives the highest activity at 300–400 °C; at 400 °C, H2 formation rate on (111) is 4.3× and 3.4× that on (100) and (110), respectively.
• Loading effect on MgO(111): For atomically dispersed Ru (<~3 wt%), TOF increases with surface Ru density, peaking at 3.43 s−1 at 400 °C (and 4.91 s−1 at 450 °C), attributed to cooperative neighboring single-atom Ru ensembles (2×/3× Ru). Ultralow-loading isolated Ru shows lower TOF (1.67 s−1). Beyond surface dispersion, formation of Ru clusters/NPs leads to a second, lower volcano with TOF up to ~2.90 s−1 at ~2 nm particle size (max B5 sites), then decreases for larger particles. Atomically dispersed ensembles on (111) outperform classical B5 nanoparticle sites in specific activity.
• DFT: On 2×Ru/MgO(111), stepwise dehydrogenation of two NH3 and N–N recombination is favorable; highest barrier corresponds to N–N coupling (~1.21 eV), consistent with experiment. H spillover from NH* to MgO(111) O sites is barrierless, whereas on (100)/(110) H migration barriers are ~1–2 eV, leading to H crowding/poisoning. Bader analysis shows spillover H increases electron density at Ru, matching AP-XPS trends. The N–N coupling involves a bimetallic µ-η2 nitrido-like intermediate akin to homogeneous systems. Breaking the last Ru–N bond and N2 desorption are more favorable on (111) than on (100)/(110).

Findings demonstrate that the polar MgO(111) support creates strong metal–support interactions with atomically dispersed Ru2+ at O-terminated hollow sites, enabling heterolytic N–H cleavage via frustrated Lewis pair behavior (Ru2+–O2−), rapid proton hopping across the surface, and electronic back-donation that weakens Ru–N bonds. This reduces hydrogen poisoning (less negative H2 reaction order), lowers apparent activation energy, and accelerates N–N recombination, especially when two Ru single atoms in close proximity cooperate to couple nitride intermediates. Compared with nonpolar MgO(100)/(110), where H binds strongly to Ru and impedes turnover, MgO(111) facilitates barrierless H spillover and reversible H2 formation. The synergy of neighboring surface single-atom Ru ensembles on MgO(111) surpasses classical nanoparticle B5 sites in specific activity and atom economy, addressing the research question of how to maximize Ru utilization and activity in ammonia decomposition.

The work establishes that atomically dispersed Ru on polar MgO(111) forms cooperative surface ensembles that outperform Ru nanoparticles with B5 sites for ammonia decomposition, owing to strong metal–support interactions that enable heterolytic NH3 activation, rapid proton spillover, and efficient N–N recombination between neighboring Ru–N species. MgO(111) delivers at least fourfold higher facet-normalized activity than MgO(100)/(110), with reduced hydrogen poisoning (γ ≈ −0.47), lower Ea (71.3 kJ mol−1), and high TOFs (up to 4.91 s−1 at 450 °C). The optimal configuration is close-proximity single-atom Ru ensembles rather than isolated single atoms or larger clusters. These insights suggest design principles for high-atom-efficiency catalysts for NH3 decomposition (and potentially synthesis), including the use of polar oxide supports to promote proton hopping and metal–support FLP behavior, and controlled creation of neighboring single-atom ensembles. Future research can explore other polar supports, tuning of Ru ensemble spacing, promoter effects, long-term stability, and integration under practical process conditions.

• The atomically dispersed Ru/MgO(111) samples, while predominantly exposing (111), still contain (100)/(110) facets; facet-specific contributions were deconvoluted statistically from multiple samples rather than measured on perfectly single-facet supports.
• DFT models exclude the explicit H2 desorption step and use slab models with specific terminations; stoichiometric and nonstoichiometric (111) gave similar Ru adsorption energies, but real surfaces may be more complex.
• AP-XPS observations were performed at 300 °C and 0.9 mbar, differing from catalytic reaction conditions; dynamic states at higher temperatures/pressures may vary.
• Long-term stability, resistance to sintering under reaction conditions, and performance under elevated pressures were not reported in detail.