Ammonia Catalyst Evolution Under Reactor Conditions Revealed by Environmental and Multimodal Electron Microscopy

Bimetallic catalysts provide new routes toward sustainable ammonia synthesis, but the structural dynamics controlling their performance under real-world conditions remain poorly understood. Here, we combine in situ gas-cell and multimodal electron microscopy to disentangle the temperature-, pressure-, and chemistry-dependent restructuring of AuRu catalysts, revealing pathways accessible only at atmospheric pressure. As synthesized, AuRu nanocatalysts are polycrystalline face-centered-cubic alloys with Au/Ru intermixing that phase-segregate into Au- and Ru-rich domains with elevated temperature (>450 °C). Increased pressure (~1 atm in 3:1, hydrogen:nitrogen) unlocks pronounced faceting and internal nanovoid formation, which systematic gas-chemistry variation identifies as hydrogen-driven. Density functional theory-based interatomic potentials show that hydrogen can amplify Au/Ru diffusion asymmetry, promoting nanovoid formation via a gas-mediated Kirkendall mechanism. Together, these results bridge the pressure gap between traditional in situ electron microscopy and benchtop ammonia reactors, enabling resolution of distinct restructuring stimuli in multicomponent systems.

💡 Research Summary

This paper addresses a critical gap in the study of ammonia synthesis catalysts by bridging the pressure difference between conventional environmental transmission electron microscopy (E‑TEM) studies (10⁻⁵–10⁻⁴ atm) and real‑world Haber‑Bosch or emerging atmospheric‑pressure reactors (up to several hundred atm). The authors develop an in‑situ gas‑cell scanning transmission electron microscopy (STEM) platform capable of operating at up to 1 atm with a 3:1 H₂:N₂ gas mixture, thereby reproducing the conditions used in recent low‑temperature, atmospheric‑pressure ammonia synthesis experiments that employ plasmonic‑metal/transition‑metal alloys such as Au‑Ru.

The study begins with a thorough multimodal characterization of as‑synthesized AuRu nanoparticles (average diameter 25 ± 6 nm). High‑angle annular dark‑field (HAADF) STEM, X‑ray energy‑dispersive spectroscopy (EDS), and 4D‑STEM reveal that the particles are polycrystalline face‑centered cubic (FCC) alloys with a lattice constant of 0.405 nm, slightly contracted relative to pure Au, indicating Ru incorporation. Elemental maps show a homogeneous Au:Ru atomic ratio of roughly 1:0.16, and low‑loss electron energy‑loss spectroscopy (EELS) detects a 2.5 eV Au‑related surface plasmon, confirming the optical signature of the alloy.

Next, the authors isolate the temperature‑driven restructuring pathway by performing stepwise vacuum annealing (20 °C min⁻¹ ramp, 10‑min holds every 50 °C). Above ~450 °C, Au and Ru begin to demix. HAADF‑STEM images display darker Ru‑rich domains that grow with temperature, while X‑EDS maps confirm spatial segregation into Au‑rich FCC and Ru‑rich hexagonal close‑packed (HCP) regions. The Ru domains average 7 ± 4 nm in size, and the overall composition after segregation remains close to the nominal Au:Ru ≈ 1:0.22, indicating near‑complete phase separation without loss of Ru to the support. The FCC→HCP transition of Ru is consistent with its thermodynamic preference and has been reported in related Au‑Ru systems.

To probe the influence of gas environment, the same particles are examined inside the gas‑cell at atmospheric pressure with a 3:1 H₂:N₂ mixture. At 1 atm, the particles develop sharply faceted {111} surfaces and, in a subset of particles, internal nanovoids that are absent under vacuum annealing. Systematic variation of the gas composition demonstrates that pure Ar or N₂ reproduces the vacuum‑only behavior, whereas pure H₂ uniquely induces nanovoid formation. This observation points to hydrogen as the chemical driver of a gas‑mediated Kirkendall effect.

Density functional theory (DFT) calculations combined with machine‑learning interatomic potentials (MLIP) are employed to quantify diffusion barriers for Au and Ru in the presence and absence of adsorbed hydrogen. The results show that hydrogen adsorption lowers the diffusion barrier for Au atoms more than for Ru atoms, creating an asymmetry that drives Au atoms outward while Ru atoms remain relatively immobile. The resulting vacancy flux leads to vacancy clustering and the observed nanovoids, a classic Kirkendall mechanism now activated by gas‑phase hydrogen at atmospheric pressure.

The authors also investigate how these structural changes affect the optical response. After vacuum‑induced segregation, low‑loss EELS line scans reveal that the Au surface plasmon at 2.5 eV persists in Au‑rich regions but is attenuated in Ru‑rich domains. Simultaneously, a new loss feature around 10–11 eV appears, localized to Ru‑rich areas, and is identified as a bulk Ru plasmon. Spatial plasmon maps thus show an anticorrelation: Au‑plasmon intensity is high where Au dominates, while Ru‑plasmon intensity is high where Ru dominates. This redistribution of plasmonic activity has direct implications for plasmon‑mediated catalysis, as the local electromagnetic field enhancement and hot‑electron generation will differ between the two domains.

In summary, the paper delivers three major insights: (1) a temperature‑driven Au‑Ru phase segregation that converts FCC Au‑rich regions and HCP Ru‑rich regions, (2) a pressure‑ and hydrogen‑dependent Kirkendall‑type nanovoid formation that only emerges at atmospheric pressure, and (3) a concomitant reshaping of the plasmonic landscape, with Au and Ru plasmons becoming spatially separated after segregation. By integrating HAADF‑STEM, 4D‑STEM, EELS, tomography, and ML‑based atomistic simulations, the authors establish a comprehensive, multimodal workflow for dissecting the intertwined effects of temperature, pressure, and gas chemistry on multicomponent catalysts. This approach not only clarifies the dynamic restructuring pathways of Au‑Ru catalysts under realistic ammonia synthesis conditions but also provides a general framework for studying other bimetallic or multimetallic systems where synergistic electronic, structural, and catalytic functions evolve simultaneously. The findings are poised to guide the rational design of next‑generation, low‑energy ammonia synthesis catalysts that exploit both catalytic and plasmonic functionalities.