Discovery of a Robust Single‐Atom Ruthenium Emission Control Catalyst

ABSTRACT

Herein, we discover a robust emission control catalyst featuring Ru single‐atom sites even undergoing thermal oxidative aging at 850°C in a 10%H₂O/10%O₂/N₂ mixture gas stream, expecting to substitute the traditional Rh catalysts by cutting ∼73% of the total cost. The stability challenges in elevated‐temperature oxidizing environments were overcome via constructing the isolated Ru atoms anchored by square‐planar coordination with four lattice oxygen in ceria and suppressing Ru atom migration toward the single Ru─O _x ─Ce catalytic centers by introducing the Zr‐rich materials, which are conducive to inhibiting the formation of undesirable N₂O by‐products during catalytic NO reduction. More importantly, the challenge of low‐temperature C─H bond activation in short‐chain alkanes on the isolated single‐atom sites was broken through by constructing the CeZrO–Ru_SA–CeO₂ three‐phase interfaces to promote the H‐spillover. The low‐cost Ru_SA–CeO₂/CZ single‐atom catalyst exhibited much better stability, lower selectivity of N₂O and NH₃ by‐products, and higher activity for CO conversion under oxygen‐lean conditions compared to commercial Rh catalysts for automotive emission control applications. This work opens new avenues for developing the new generation of low‐cost, robust emission control catalysts in the future.

The stability challenges in elevated‐temperature oxidizing environments were overcome by constructing the isolated Ru atoms anchored on ceria and suppressing Ru atom migration toward the single Ru─O_x─Ce catalytic centers via the gaseous RuO₄ trapping by Zr‐rich materials, expecting to substitute the traditional Rh catalysts by cutting ∼73% of the total cost.

1. Introduction

The automotive industry consumed 60.5% of the global total demand of Pt, Pd, and Rh to meet the emission control regulations in the past 20 years (Figure S1). The rapid development of the automotive industry triggered the supply‐demand contradictions and price fluctuations. The dependence of automotive catalysts on Rh was particularly astonishing, consuming 86.0% in average, and reaching a maximum of 94.3% in 2021 (Figure S1c), which makes the drastic fluctuations in Rh prices caused by its dynamic demand (Figure S2). Ways to achieve a low‐cost and reliable emission control catalyst must be developed in terms of the substitution strategy of those extremely economically unfavorable elements, such as Rh. However, the complete substitution of Rh by other elements in automotive catalysts seems to be a long‐standing challenge to simultaneously convert carbon monoxide, unburned hydrocarbons, and nitrogen oxides into water, carbon dioxide, and nitrogen, owing to the specific stability and selectivity issues [1, 2, 3, 4]. The traditional supported Rh catalysts for three‐way catalysis applications are easily deactivated by sintering the active components and generating the rhodium aluminate (RhAlO _x ) structure [5, 6], which requires the use of ultra‐stable Ce‐based oxygen storage materials to avoid the Rh‐Al₂O₃ interfaces.

The emergence of single‐atom catalysts provides the possibility for significantly reducing the Rh loadings in automotive catalysts while maintaining the same or better catalytic activity and stability [3, 7, 8]. Unfortunately, the undesirable NH₃ by‐product is readily formed on the atomically dispersed Rh active sites during catalytic NO reduction [9, 10, 11]. Ru was culled out of the list of candidates for the vital active metals in automotive catalysts long ago due to forming volatile oxides in an elevated‐temperature oxidizing environment, retarding its application in automotive catalysts, although it is much cheaper than Rh (Figure S2) [12]. In recent years, some Ru single‐atom catalysts have been developed to apply in emission control [13, 14, 15], but their thermal stability issue in an elevated‐temperature (>800°C) oxidizing stream with high oxygen concentrations has not made a significant breakthrough, which directly hinders commercial viability in automotive three‐way catalysis.

Herein, we disentangle the selectivity‐stability trade‐off in catalytic NO reduction by developing a Ru single‐atom catalyst with the isolated Ru_penta atoms anchored on ceria surfaces and proposing a gaseous RuO₄ trapping strategy to suppress the agglomeration of Ru atoms in an elevated‐temperature oxidizing environment. Our experimental results demonstrate that the single‐atom ruthenium emission control catalyst reduced 86% of N₂O production during the dynamic oxidation/reduction catalysis and exhibited a much better stability, compared to that of commercial Rh catalysts for automotive emission control applications.

2. Results and Discussion

2.1. Catalyst Structures

The Ru─O─Ce(OH) _z precursor with 18.5% of Ce³⁺ species was synthesized via a like condensation reaction (Figures 1a and S3). XPS spectrum shows that the chlorine element was not detected in as‐synthesized Ru─O─Ce(OH) _z powder (Figure S4a), confirming the removal of Cl^– ligands on Ru atoms. The Ru─Cl bond could be broken by reacting with activated hydroxyls on the surface of [Ce(OH) _y ] ^{ζ +} colloids in a hydrothermal environment to release H⁺ and Cl^– ions into the liquid phase, which leads to a significant decrease in pH of filtrate as described in the Methods section and the existence of residual Cl in the corresponding evaporated residue (Figure S4b). By contrast, residual Cl was detected and still bonded to Ru atoms in the Ru/CeO₂ catalyst synthesized by the conventional wet impregnation method (Figure S5). To further confirm this, we selected CO as a probe molecule to adsorb onto the Ru sites at 30°C, showing that it maintains the RuCl _n structures in the Ru/CeO₂ catalyst due to the determination of Ru(CO) _m Cl _n species after CO adsorption (Figure S6 and Table S1) [16].

FIGURE 1.

Schematic diagram illustrating the fabrication processes and structure characterizations of the Ru_SA─CeO₂ single‐atom catalyst. (a) Positively charged [Ce(OH) _y ] ^{ζ +} colloids were synthesized by reacting Ce³⁺ cations with aqueous ammonia, followed by the hydrothermal reaction with hydrated [RuCl _x ] ^δ– species at 120°C for 24 h to obtain Ru─O─Ce(OH) _z precursor, and single Ru atoms anchored in the ceria lattice were achieved by calcination at 600°C for 2 h in nitrogen stream with a heating rate of 2°C/min. (b) Overlapping of the HAADF‐STEM image with the EDS maps of the Ru_SA─CeO₂ single‐atom catalyst. (c) In situ DRIFTS spectra collected from CO adsorption on the Ru_SA─CeO₂ (red) and Ru_NP─CeO₂ (navy) samples after purging with N₂ for 10 min at 30°C. No bridged‐CO species (1843 cm⁻¹) was detected on the surface of CO‐adsorbed Ru_SA─CeO₂ catalyst. (d) EXAFS spectra of Ru K‐edge in the Ru_SA─CeO₂ catalyst and references. (e–g) Wavelet transform of the k ²‐weighted FT‐EXAFS spectra at the Ru K‐edge of Ru foil (e), RuO₂ (f), and Ru_SA─CeO₂ catalyst (g). (h,i) Front view of the isolated Ru_penta sites (h) and Ru─O_v sites (i) anchored on the considered CeO₂(100) surfaces. (j) The relative energy of Ru species as a function of the oxygen chemical potential of O, μ(O). Color code: O–red; Ce–yellow; Ru–gray.

Calcination of the collected Ru─O─Ce(OH) _z powder at 600°C in a N₂ stream for 2 h increases the surface Ce³⁺/Ce percentage from 18.5% to 21.9% (Figures S3) and produces the single Ru atoms anchored in ceria lattice with the Ru─O─Ce bridges (Figure S7). The Raman spectra of Ru_SA–CeO₂ showed a F_2g vibration mode (446 cm⁻¹) related to the octahedral local symmetry of CeO₂ nanocrystals and two clear bands at 683 and 957 cm⁻¹ assigned to the asymmetric Ru─O─Ce structure, suggesting that the cationic Ru ^{n +} species were successfully introduced into CeO₂ [14, 17, 18]. As shown in Figure S8, line‐scan profiles were collected across different atomic columns in the Ru_SA─CeO₂ catalyst. Along each scan line, the dominant high‐intensity peaks correspond to the Ce columns (Z = 58) in the CeO₂ lattice. Notably, several intermediate‐intensity peaks were consistently observed between these Ce‐related signals. The intensities of these intermediate peaks fall clearly below those of Ce but well above the background noise, indicating the presence of individual Ru (Z = 44) atoms due to the Z‐dependent intensity difference between Ru and Ce [19]. The reproducible appearance of these distinct intermediate peaks across multiple line scans (A1–A2, B1–B2, and C1–C2) verifies the presence of atomically dispersed Ru sites on the CeO₂ support. The atomic dispersion of Ru can be further confirmed by the EDS line‐scanning profile of the Ru element, as shown in Figure S9. The EDS maps (Figure 1b) confirm that Ru species are uniformly distributed on the CeO₂ support without the observation of Ru nanoparticles, excluding the presence of Ru NPs in the Ru_SA─CeO₂ catalyst [20, 21].

DRIFTS spectrum of CO adsorbed on the Ru_SA‐CeO₂ single‐atom catalyst (Figure 1c) showed four strong peaks centered at 2120, 2056, 1975, and 1935 cm⁻¹, which can be assigned to the surface carbonyls with a total CO adsorption capacity of 181 µmol/g_cat. at 30°C (Tables S1 and S2). The absorption band at 2120 cm⁻¹ corresponds to C─O stretching vibrations of CO linearly bound Ce³⁺ sites and tricarbonyl species bonded to the atomically dispersed Ru ^{n +} sites [22, 23]. The vibrational frequencies at 2056 cm⁻¹ can be ascribed to the C─O vibrations of geminal dicarbonyls adsorbed on isolated cationic Ru [24, 25]. The other peaks centered at 1975 and 1935 cm⁻¹ are closely related to the linearly adsorbed CO on Ru atoms with high and low coordination at the Ru─CeO₂ interfaces as Ce─O‐Ru─CO and Ce‐vacancy‐Ru─CO, respectively [25, 26]. In the CO‐DRIFTS spectrum of Ru_SA─CeO₂ (Figure 1c), there are no observations of the C─O stretching vibrations at 2040, 1956, and 1843 cm⁻¹ assigned to the linear and bridged CO adsorbed on metallic atoms of Ru clusters and nanoparticles [22, 27]. Figure S10 shows the HAADF‐STEM characterization of Ru NPs on a CeO₂. Ru nanoparticles were observable on the CeO₂ surface in the low‐magnification HRTEM image; a high‐magnification view of the red‐boxed area on the Figure S10a further reveals the atomic‐level lattice structure of the Ru particle (Figure S10b); furthermore, lattice spacing analysis of the yellow‐boxed region, using intensity profiles and 3D atomic reconstruction, yields an interplanar spacing of 0.22 nm, which matches the characteristic spacing of the (101) plane of metallic Ru (Figure S10c). Additionally, the CO vibrational frequencies of multi‐carbonyls bonded to positively charged Ru ^{n +} sites shift towards higher wavenumber compared to that of the Ru_NP─CeO₂ sample, showing the distinguishing features of mononuclear Ru(CO) _m species [28].

The clear N─O stretching vibrations in Ce─O─Ru−NO (1847 cm⁻¹) and Ce‐vacancy‐Ru─NO (1764 cm⁻¹) structures were also detected after NO adsorption at 400°C, and the NO vibrational frequency of Ru─NO (1820 cm⁻¹) was lower than that of Ru/CeO₂ sample (1865 cm⁻¹) due to the lack of Cl ligands with a higher electronegativity compared to Ru (Figure S11) [29]. X‐ray absorption spectra were employed to insight the coordination environments of Ru species. Ru K‐edge XANES spectrum of the Ru_SA‐CeO₂ single‐atom catalyst (Figure S12) indicates that the Ru is cationic due to its oxidation state. In addition, its EXAFS spectrum (Figure 1d) shows a single major peak centered at 1.41 Å associated with the scattering contributions of the Ru─O bond compared to that of the Ru_NP─CeO₂ nanostructured catalyst (Figure S13) [14], providing evidence of atomically dispersed Ru. Further EXAFS curve fitting results (Table S3) give the coordination number of 5.4 ± 1.8 and the average Ru─O bond length of 1.98 Å. Wavelet transform of the k ²‐weighted FT‐EXAFS spectra at the Ru K‐edge suggested that isolated Ru atoms were anchored in the ceria lattice via the Ru─O─Ce bridges (Figure 1e–g), consistent with the Raman and NO‐DRIFTS results.

As mentioned above, we concluded that the as‐synthesized Ru_SA‐CeO₂ single‐atom catalyst features the isolated Ru‐O and Ru‐O_v sites (Figure 1h,i). The relative energies of single Ru atoms with different oxygen‐bridged structures (Figure S14) as a function of the oxygen chemical potential were calculated to understand the stability of the isolated Ru sites. As shown in Figure 1h, the isolated Ru_penta site, anchored by square planar coordination with four lattice oxygen and one vertical oxygen adsorption, is the most stable one among the considered coordination structures of single Ru atoms. In addition, the isolated Ru_penta site shows a lower relative energy, compared to that of gaseous RuO₄ at 1000 K and an O₂ partial pressure of 1 atm. These results suggest that the pentacoordinate Ru single site can be denoted as O─Ru─O₄─Ce₄‐d, and that its enhanced stability effectively suppresses the formation of gaseous RuO₄. In the case of the Ru_SA─CeO₂ catalyst, gaseous RuO₄ started to be detected during the O₂‐TPO‐MS experiment when the temperature of the 10%O₂/N₂ stream exceeded 780°C (Figure S15a), which perfectly validates the theoretical calculation results. Consequently, the infrared peak centered at 2120 cm⁻¹ in Figure 1c can be accurately assigned to the C─O stretching vibrations of tricarbonyl species from the O─Ru(CO)₃─O₄─Ce₄‐d structure, well explaining why the wavenumber is higher than that of tricarbonyl species adsorbed on Ru nanoparticles.

The average grain size of ceria was 10.1 nm in the Ru_SA─CeO₂ single‐atom catalyst, calculated by the Scherrer equation using x‐ray diffraction (XRD) line‐width broadening data assigned to the CeO₂ (111) plane (Figure S15b), smaller than 14.6 nm of pure CeO₂ synthesized by the same procedure (Table S2). The results suggest that the condensation reaction of [RuCl _x ] ^δ– species with [Ce(OH) _y ] ^{ζ +} colloids suppressed the further agglomeration of positively charged [Ce(OH) _y ] ^{ζ +} colloids during the hydrothermal synthesis process, thereby exhibiting nearly twice the specific surface area and total pore volume as as‐synthesized CeO₂. In summary, all of the abovementioned findings fully demonstrate that the thermodynamically stable O─Ru─O₄─Ce₄‐d catalytic centers were successfully fabricated by anchoring Ru single‐atom sites in the surface lattice of ceria through the condensation reaction of [RuCl _x ] ^δ– species with [Ce(OH) _y ] ^{ζ +} colloids in a mild hydrothermal environment and the thermally driven oxygen vacancy generation under inert atmosphere, respectively.

2.2. Catalytic Performance and Elevated‐Temperature Stability

Catalytic reduction of NO by CO has been identified as a vital reaction for the nitrogen oxide elimination in environmental catalysis [13]. Curves of CO and NO conversion over the Ru_SA─CeO₂ single‐atom catalyst under the dry and wet conditions provide clear evidence of the water‐assisted low‐temperature activation, which can be verified by the apparent activation energy (E _a) decline from 49 to 41 kJ∙mol⁻¹ and the significant increments in NH₃ and CO₂ formation by introducing 10% water vapor (Figure S16 and Table S4). NH₃ formation under dry conditions can be attributed to the surface hydroxyls mediated NO reduction in CeO₂ crystallites [30]. NH₃ yields were significantly increased when 10% water vapor was introduced, resulting from the additional supplementation of activated hydrogen species via the water–gas shift (WGS) reaction. Low‐temperature activity for converting CO into CO₂ in the WGS region followed the order of Ru_SA─CeO₂> Rh/CeO₂> Pd/CeO₂ (Figure S17a), which agrees well with precious report [31]. In addition, Ru_SA─CeO₂ single‐atom catalyst exhibited the superior ability to inhibit the formation of undesirable N₂O byproduct in the emission control region and NH₃ by‐product at 420°C under rich conditions (Figures S17b and S18), compared to that of Rh/CeO₂ and Pd/CeO₂ catalysts.

The kinetics were investigated in the Arrhenius‐type plot for the model reaction of catalytic NO reduction by CO in the presence of 10% water vapor (Figures S19 and S20 and Table S4). The E _a value of NO reduction over the Ru_SA─CeO₂ single‐atom catalyst is slightly lower than that of the Rh─F catalyst (46 kJ∙mol⁻¹), and close to half of E _a (73 kJ·mol⁻¹) for the Ru/CeO₂ supported catalyst. As shown in Figure S21, the undesirable NH₃ and N₂O byproducts are more readily formed over the Rh─F catalyst, showing a N₂O selectivity of 43.7% at 180°C and an NH₃ selectivity of 61.9% at 420°C, which is much higher than that of Ru catalyst. Interestingly, the N₂O selectivity at 180°C for Ru/CeO₂ exceeded twice as much as Ru_SA─CeO₂, suggesting that the Ru─O _x ─Ce catalytic centers can against the formation of N₂O by‐product. To evaluate the transient N₂O formed on Ru single‐atom sites, we developed a robust Ru_SA─CeO₂/CZ single‐atom automotive catalyst by physically mixing as‐synthesized Ru_SA─CeO₂ single‐atom catalyst with commercially available Zr‐rich Ce‐based composite oxides. As for Ru_SA─CeO₂/CZ single‐atom catalyst, the maximum concentration of N₂O byproduct was only 7 ppm centered at 217°C in the light‐off experiment of the simulated oxidation/reduction catalysis, which reduces by about 86% comparing with the maximum N₂O concentration (49 ppm) formed over the Rh─F catalyst (Figure S22).

The stability challenges of the new materials concepts, including single‐atom catalysts, still need to be overcome for revolutionizing the current emission control systems of automobiles [7]. The cyclic stability test of oxidation/reduction catalysis over the Ru_SA─CeO₂/CZ single‐atom catalyst shows that the single‐atom Ru catalyst did not deactivate during the heating/cooling cycles (Figures S23–S26). The mean light‐off temperatures for NO, CO, C₃H₆, and C₃H₈ conversion were 226°C, 213°C, 240°C, and 402°C (Figure S27), respectively, which is superior to the results over the single‐atom Pt, Pd, and Rh catalysts, as reported by Lee group [8]. Gasoline vehicle exhaust generally contains high concentrations of oxygen gas under the deceleration driving conditions over the World‐wide harmonized Light duty Test Cycle (WLTC) according to our previous report (Figure S28) [32], posing a significant threat to the stability of Ru‐based automotive catalysts. Figure S29 shows that the inlet exhaust temperature is below 700°C and decreases 150°C∼240°C caused by deceleration driving at ultra‐high speed. Consequently, a 10‐h thermal oxidative stability test at 850°C in 2 L/min of mixture gas stream containing 10%H₂O and 10%O₂ was employed to evaluate the durability of Ru_SA─CeO₂/CZ single‐atom catalyst, considering the mode of application in the under‐floor catalyst systems. In addition, 10‐h hydrothermal aging was also performed at 900⁰C in 10% H₂O/N₂ flow to further confirm the reliability of the thermal oxidative aging mode.

As can be seen in Figures 2a and S30, thermal oxidative aging at 850⁰C led to a significant decrease in activity for converting CO and NO into CO₂ and N₂ in the emission control and WGS region. Ru_SA─CeO₂/CZ single‐atom catalyst showed an absolute superiority of catalytic activity for the model reaction of catalytic NO reduction by CO after thermal oxidative aging by comparison with the Rh‐TOA catalyst (Figures 2a,b and S31). Light‐off temperatures for NO and CO conversion over the Ru_SA─CeO₂/CZ‐TOA catalyst were 144°C and 207°C, respectively, increasing 2°C and 16°C caused by thermal oxidative aging (Figures 2b and S32). By contrast, thermal oxidative aging at 850°C resulted in a dramatic decline of catalytic activity for NO reduction by CO and WGS reaction according to the curve of CO conversion over the Rh‐TOA catalyst (Figures 2a and S30). Ru_SA‐CeO₂/CZ single‐atom catalyst remained the similar ability to remove CO contaminant via WGS reaction at 420°C after thermal oxidative aging, while the commercial Rh catalyst suffered a significant decline in WGS activity (Figure 2a,c). In addition, Ru_SA─CeO₂/CZ single‐atom catalyst showed a superior ability to inhibit the formation of undesirable N₂O byproduct, compared to that of commercial Rh catalysts (Figure S33). These findings provide a possibility to Ru_SA─CeO₂/CZ single‐atom catalyst, and suggest that achieving the efficient purification of CO contaminant will not be strongly dependent on oxygen storage capacity of Ru_SA─CeO₂/CZ, which is one of the significant features compared to the traditional automotive catalysts.

FIGURE 2.

Stability of the Ru_SA─CeO₂/CZ single‐atom catalyst in elevated‐temperature oxidizing environment. (a,b) Curves of CO conversion during NO reduction by CO over the commercial Rh catalysts under wet conditions. (c,d) CO conversion during the steady‐state oxidation/reduction reaction over the fresh and aged Ru_SA─CeO₂/CZ (c) and commercial Rh (d) catalysts at 420°C as a function of the oxygen excess coefficient. (e–h) Curves of NO and C₃H₆ conversion during the dynamic oxidation/reduction catalysis over the fresh and aged Ru_SA─CeO₂/CZ and commercial Rh catalysts.

Dynamic oxidation/reduction catalysis behaviors were also evaluated to further identify the superiority of durability. Ru_SA─CeO₂/CZ single‐atom catalyst maintained an equivalent activity for NO reduction and C₃H₆ oxidation and showed a slight activity loss for CO oxidation after hydrothermal aging at 850°C for 10 h, while hydrothermal aging induced a significant loss of activity for NO, CO, C₃H₆, and C₃H₈ conversion over the commercial Rh catalyst (Figures 2e–h and S34). In addition, Ru_SA─CeO₂/CZ single‐atom catalyst exhibited a better inhibition ability to convert NO towards undesirable N₂O and NH₃ by‐products than commercial Rh catalyst (Figures S35 and S36). These results illustrate that the durability of Ru_SA─CeO₂/CZ single‐atom catalyst far exceeds that of the commercial Rh catalyst.

2.3. Stabilization Mechanisms of Ru Single‐Atom Sites

The in‐situ CO‐DRIFTS spectra of Ru_NP─CeO₂ and Ru/CeO₂‐TOA catalyst showed a large broad band centered at 1843 cm⁻¹, which can be assigned to the C–O vibrations of bridged‐carbonyl species adsorbed on Ru nanoparticles (Figure S37a,b). The Ru_SA─CeO₂ single‐atom catalyst also suffered a significant sintering via the migration of volatile Ru‐oxide molecules to form Ru‐oxide nanoparticles, resulting from the detection of C─O vibration at 1856 cm⁻¹ attributed to the bridged Ru₂─CO structure on the surface of RuO₂‐like particles (Figures S15a, S37c, and S38) [26, 33]. As shown in Figure S15a, the detected MS signals from gaseous RuO₄ were much higher than those of RuO₃ in an elevated‐temperature oxidizing environment, which provides the possibility for Ru atom migration by forming the volatile RuO₄ molecule. To further confirm the formation of Ru nanoparticles by the volatile Ru‐oxide molecule migration, we employed the as‐synthesized CeO₂ powder to trap the volatile Ru‐oxide molecules from the Ru_SA─CeO₂ single‐atom catalyst in the 10%O₂/10%H₂O/N₂ stream at 850°C for 10 h, thereby achieving the Ru_TRAP─CeO₂‐TOA sample. The observation of a clear peak centered at 1835 cm⁻¹ in the in situ CO‐DRIFTS spectrum of the Ru_TRAP─CeO₂‐TOA sample suggests that the volatile Ru‐oxide molecules were trapped by CeO₂ to form Ru nanoparticles (Figure 3a). These results fully indicate that there is no possibility to maintain the isolated Ru single‐atom sites via atom‐trapping by ceria in an elevated‐temperature (∼850°C) oxidizing stream with high oxygen concentrations. In Figure 3a, the bands at 2042 and 1971 cm⁻¹ can be attributed to the asymmetric and symmetric vibrations of dicarbonyl species bound to the Ru atoms trapped by CeO₂ [24]; the band at 1835 cm⁻¹ was attributed to bridged‐carbonyls on Ru nanoclusters/nanoparticles, indicating the insufficient anchoring capability of atomic dispersed Ru on CeO₂ under this operation condition (Ru_TRAP─CeO₂‐TOA). By contrast, no IR band assigned to bridged‐carbonyls was observed in the in‐situ CO‐DRIFTS spectrum of the Ru_TRAP─ZrO₂‐TOA sample (Figure 3b), indicating that ZrO₂ helps to maintain Ru in a single‐atom state under elevated‐temperature oxidizing conditions. Thus, the bands at 2064 and 1900 cm⁻¹ should be closely related to the C─O vibrations from the ZrO₂─Ru_SA ²⁺‐(CO)₃ structure [34].

FIGURE 3.

Stabilization mechanisms of the active Ru sites in elevated‐temperature oxidizing environments. (a–c) In‐situ DRIFTS spectra collected from CO adsorption on the Ru_TRAP─CeO₂‐TOA (a), Ru_TRAP─ZrO₂‐TOA (b), and Ru_TRAP─CZ‐TOA (c) samples at 30°C. (d) H₂‐TPR profiles of the CZ‐TOA and Ru_TRAP‐CZ‐TOA samples. (e) In situ DRIFTS spectra collected from CO adsorption on the fresh and aged catalysts after purging with N₂ for 10 min at 30°C. No bridged‐CO species was detected on the surface of CO‐adsorbed Ru_SA─CeO₂/CZ‐TOA and Ru_SA─CeO₂/CZ‐HA single‐atom catalysts. (f) H₂‐TPR profiles of the Ru_SA─CeO₂‐TOA, Ru_SA─CeO₂/CZ‐TOA, and Ru_TRAP─CZ‐TOA catalysts. (g) The agglomeration of Ru atoms at elevated temperatures was significantly suppressed by ball‐milling mixing with Zr‐rich Ce‐based composite oxides, due to the atom trapping effect during the thermal oxidative aging at 850°C.

No clear IR band assigned to bridged‐carbonyls can be observed in Figure 3c, and its dominant IR peak appeared at 2044 cm⁻¹, which can be reasonably ascribed to the asymmetric C‐O vibration of CeO₂‐Ru_SA–(CO)₂ structure. These results suggest that the CeO₂ sites favor to trap the gaseous RuO _x molecules, and the existence of ZrO₂ sites can effectively prevent the agglomeration of Ru atoms in an elevated‐temperature oxidizing environment. This viewpoint can be further confirmed by the observation of significant H‐spillover at CeZrO─Ru_SA in the Ru_TRAP‐CZ‐TOA sample (Figures 3d and S39). Thus, we used Zr‐rich Ce‐based composite oxides in ball‐milling to obtain the Ru_SA─CeO₂/CZ single‐atom catalyst with a slight decline in specific surface area and total pore volume (Table S5). Any Ru atoms on CZ oxides and crystal structure transformation were not observed in Ru_SA─CeO₂/CZ‐F catalyst (Figures S40–S42). Ru_SA‐CeO₂ catalyst consumed much lower hydrogen around 167°C assigned to the reduction of surface oxygen species near to Ru atoms [18], compared to that of Ru_NP─CeO₂ catalyst (Figures S43 and S44). The poor hydrogen consumption at low temperatures suggests the strong metal–support interaction in Ru_SA‐CeO₂ catalyst. However, the H₂‐TPR profile of the Ru_SA─CeO₂/CZ‐F single‐atom catalyst appeared a strong hydrogen consumption peak centered at 169°C, indicating that the generation of CeZrO/Ru_SA─CeO₂ interfaces significantly promotes H‐spillover (Figure S44). The C─O stretching vibrations of Ce─O─Ru–(CO)₂, Ce─O─Ru─CO, and Ce‐vacancy‐Ru─CO species shifted towards higher wavenumbers, reaching 2073, 1999, and 1968 cm⁻¹, respectively, and an additional peaks appeared at 2020 cm⁻¹ assigned to linear carbonyl on Ru sites with high coordination [14], caused by ball‐milling mixing with CZ oxides (Figure 3e).

Interestingly, unlike the Ru_SA─CeO₂ catalyst, Ru is still atomically dispersed in the Ru_SA─CeO₂/CZ‐TOA catalyst even undergoing the thermal oxidative aging at 850°C for 10 h, although 64% of the specific surface area and 27% of total pore volume were declined, which is confirmed by no observation of the IR band attributed to bridged‐carbonyl in CO‐DRIFTS spectrum (Figure 3e). No IR band at 1860∼1830 cm⁻¹ was detected in the in‐situ CO‐DRIFTS spectra of Ru_SA─CeO₂/CZ‐HA and Ru_SA─CeO₂‐HA (Figures 3a and S45a). In addition, any IR signal assigned to the C─O stretching vibrations of carbonyls bonded to Ru atoms was not observed from CO‐DRIFTS spectra of the Ru_TRAP‐CZ‐HA sample (Figure S45b). These results illustrate that isolated Ru atoms anchored on ceria are ultra‐stable without the formation of volatile Ru‐oxide molecules in elevated‐temperature hydrothermal environments. Ru atoms were uniformly located on the CeZrO support in the Ru_SA─CeO₂/CZ‐TOA sample (Figure S46), providing evidence of the Ru atom migration into Zr‐rich Ce‐based composite oxides. The H‐spillover promoted by the generation of CeZrO‐Ru_SA─CeO₂ three‐phase interfaces (Figure 3f) allows the low‐temperature C─H bond activation of short‐chain alkanes, explaining why the C₃H₈ conversion was enhanced after thermal oxidative aging (Figure S34d). This finding overturns the insights that C─H bonds are difficult to activate by isolated single‐atom sites during the catalytic conversion of short‐chain alkanes [35]. The mentioned Ru atom trapping effect of Zr‐rich materials can protect the single Ru─O _x ─Ce catalytic centers from thermally induced sintering to avoid the formation of Ru clusters and maintain its intrinsic activity for catalytic NO by CO (Figure S47). Figure S48 suggests that the constructed single Ru─O _x ─Ce catalytic centers and the introduction of Zr‐rich materials exhibit an excellent antioxidation performance to markedly reduce the release of the volatile RuO₄ species. Thus, as displayed in Figure 3g, stabilization mechanisms of Ru single‐atom sites can be summarized as the gaseous RuO₄ trapping by Zr‐rich materials to suppress Ru atom migration towards the single Ru─O _x ─Ce catalytic centers in an elevated‐temperature (∼850°C) oxidizing stream.

2.4. Substitution Potential of Ru for Rh in Under‐Floor Emission Control Catalysts

As for the gasoline‐fueled vehicles, the maximum exhaust temperature in the under‐floor positions is within 730°C and generally about 100°C lower than that of the close‐couple positions over the typical driving cycles [36, 37], suggesting that an aging temperature of 850°C is enough to evaluate the durability for application as an under‐floor emission control catalyst. However, some potential harsh driving conditions can make it possible to undergo the transient overtemperature. Thus, a higher temperature oxidative aging at 1000°C was also employed to further evaluate the superiority in durability of Ru_SA─CeO₂/CZ catalyst. Figure 4a,b shows that Ru_SA‐CeO₂/CZ‐HTOA catalyst exhibited a significantly better low‐temperature activity for the catalytic NO reduction by CO than that of Rh‐HTOA, shifting towards lower temperatures by ∼140°C of light‐off temperature for NO conversion and maintaining a light‐off temperature of 253°C for CO conversion. In particular, the Ru_SA─CeO₂/CZ‐HTOA catalyst exhibited a superior ability to convert CO in the WGS region, compared to the commercial Rh‐HTOA catalyst. The steady‐state CO conversion plots (Figure 4c) can provide further evidence of its superiority in WGS activity. In addition, the lower steady‐state NH₃ selectivity over the Ru_SA‐CeO₂/CZ‐HTOA catalyst at 420°C (Figure 4d) indicates that substituting Rh by Ru in under‐floor emission control catalysts can help to reduce the NH₃ secondary pollutant emission, which is crucial for the next stage of automotive emission control [38].

FIGURE 4.

Comparison of the stability at 1000°C between Ru_SA‐CeO₂/CZ and commercial Rh catalysts. (a,b) Curves of NO (a) and CO (b) conversion during the catalytic NO reduction by CO under wet conditions. (c,d) CO conversion (c) and NH₃ selectivity (d) during the steady‐state oxidation/reduction reaction at 420°C as a function of the oxygen excess coefficient. (e–h) Curves of CO (e), C₃H₆ (f), NO (g) conversion, and N₂O byproduct (h) formation during the dynamic oxidation/reduction catalysis.

As reported by Jang et al. [36], an under‐floor emission control catalyst predominantly operates at around 420°C in the aftertreatment system of gasoline‐fueled engines under normal driving conditions, while it requires a longer time to effectively operate during the cold‐start period than that of a close‐couple catalyst. By comparison with the commercial Rh‐HTOA catalyst, Ru_SA─CeO₂/CZ‐HTOA showed a better light‐off behavior during the dynamic oxidation/reduction catalysis (Figure 4e–g) and a much more excellent ability to inhibit the formation of N₂O by‐product at the temperature range of 300°C–400°C (Figure 4h). This provides a new approach to address the issue of a significant increase in N₂O secondary pollutant emissions under low‐speed driving conditions caused by the degradation of Rh‐containing catalysts (reported by Zheng and her co‐workers [39]). Besides, the price of Rh was commonly ∼10 times and even reached up to ∼80 times as much as Ru in the past 20 years (Figure S49). As mentioned above, it is technically and commercially viable to substitute Rh by Ru in an under‐floor emission control catalyst.

3. Conclusion

In summary, it is viable to achieve the complete substitution of Rh by Ru in a three‐way catalyst by overcoming the stability challenges in elevated‐temperature oxidizing environments via constructing the isolated Ru_penta atoms anchored by square planar coordination with four lattice oxygen on ceria and combining with the gaseous RuO₄ trapping effect of Zr‐rich materials. The Ru_SA─CeO₂/CZ single‐atom catalyst is robust: (1) Ru is maintained atomically dispersed, and there is no catalytic activity loss after thermal oxidative aging at 850°C in the 10%H₂O/10%O₂/N₂ mixture gas stream, (2) it shows much better superiority in durability and selectivity even undergoing thermal oxidative aging at 1000°C compared to a commercial Rh catalyst. We also made a breakthrough in the challenge of low‐temperature C─H activation on the isolated single‐atom sites during the catalytic conversion of short‐chain alkanes by constructing the CeZrO─Ru_SA─CeO₂ three‐phase interfaces to promote the H‐spillover. Compared to commercial Rh catalysts, the Ru_SA─CeO₂/CZ single‐atom catalyst can cut the total cost by ∼73% (Table S6) with much better stability, lower selectivity of N₂O and NH₃ byproducts, and higher activity for CO conversion under oxygen‐lean conditions, indicating that it is very suitable as a new generation of three‐way catalyst. As a consequence, we believe that Ru will be listed as the candidate for the vital active metals in the next generation of automotive catalysts that are more reliable and inexpensive.

Author Contributions

Chengxiong Wang: conceptualization, data curation, formal analysis, funding acquisition, investigation, visualization, writing – original draft. Jin‐Cheng Liu: methodology, visualization, writing – review and editing. Zhi Li: project administration, writing – review and editing. Xinyu Liu: investigation, data curation. Xiao Liang: data curation, formal analysis. Yang Li: methodology, resources. Mengyun Wang: data curation, formal analysis. Yiming Zhao: investigation, validation. Guangxin Liu: investigation, data curation. Shuangchao Yao: investigation. Rong Wang: investigation. Chen Chen: methodology, resources. Dingsheng Wang: methodology, resources. Qing Peng: methodology, resources.Yunkun Zhao: methodology, project administration, supervision, writing – review and editing. Yadong Li: conceptualization, funding acquisition, project administration, supervision, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie73302‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.