Cooperative Redox Catalysis of N₂O Decomposition by Short‐Range RhO _x and CeO _x Anchored to Co₃O₄

ABSTRACT

Hybrid materials based upon mixed metal oxides provide an effective strategy to boost catalytic performance through redox‐induced interfacial engineering. Herein, we present an efficient catalyst toward N₂O decomposition containing short‐range ordered RhO _x and CeO _x species anchored to the Co₃O₄ spinel. The activity enhancement relative to Co₃O₄ is non‐additive with respect to the RhO _x /Co₃O₄ and CeO _x /Co₃O₄ interfaces and is uniquely linked to the modulation of the Co and Rh oxidation states at a synergistic Rh–O–Co–O–Ce interface. This ensemble provides increased redox flexibility and facilitated lattice oxygen activation, resulting in markedly enhanced N₂O decomposition. Kinetic analysis and operando modulated excitation x‐ray absorption spectroscopy (ME‐XAS) provide direct mechanistic evidence that N₂O decomposition proceeds via a multi‐site, cooperative Mars‐van Krevelen pathway, in which Co₃O₄–CeO₂ acts as an oxygen reservoir enabling the redox cycling of RhO _x species. This study highlights the ability of mixed metal oxide catalysts to exploit interfacial effects to achieve improved activity in redox‐mediated reactions and offers a rationale for the engineering of metal oxide–metal oxide catalysts guided by operando spectroscopy.

A short‐range Rh–O–Co–O–Ce interfacial moiety drives low‐temperature N₂O decomposition via a cooperative redox cycle. Operando modulated‐excitation XAS with phase‐sensitive detection (ME‐XAS + PSD) tracks rapid, reversible Rh oxidation coupled to Co reduction. Combined kinetics and DFT rationalize the activity trends and identify the Rh–O–Co interface as a key descriptor.

1. Introduction

Noble metals (NM) are scarce and costly, yet they present unique catalytic properties across a wide range of catalytically driven chemical transformations [1, 2]. To improve their utilization, large research efforts have been devoted to the design of catalysts, evolving from nanoparticles to short‐range ordered structures and, ultimately, to single‐atom catalysts [3]. While these approaches improve atom efficiency and often enhance activity, they also pose challenges in terms of stability, electronic control, and unambiguous identification of active sites [4].

In this context, metal oxide‐metal oxide hybrid materials have emerged as promising support materials for noble metals to address these limitations, because these systems often display markedly interfacial metal–support effects [5, 6, 7]. These interactions can effectively impact catalytic performance [8, 9, 10, 11], while simultaneously providing a platform for catalysts engineering, as they can stabilize noble metal atoms anchored at the oxide interface [12, 13]. These effects have been extensively rationalized in terms of tuning of the metal d‐band centers [5, 14].

Cobalt oxide (Co₃O₄) is a particularly attractive metal oxide support owing to its mixed‐valence (Co²⁺, Co³⁺) spinel structure, intrinsic redox activity, and tunable d‐orbital occupancy [15, 16]. Noble metals (Pt [11, 12, 17, 18], Pd [12], Ir [19], Ag [20], Au [21, 22], Rh [23, 24]) supported on Co₃O₄ have drawn increasing attention due to the tunable NM/Co₃O₄ interface, and have shown promise in reactions such as CO oxidation [11, 17, 21, 22], H₃NBH₃ hydrolysis [18], and VOC oxidation reactions (e.g., toluene, benzene, etc.) [25]. Moreover, short‐range‐ordered NM species anchored on Co₃O₄ are highly effective motifs for oxygen evolution reactions [26, 27, 28]. Beyond interfacial effects, bulk substitution strategies (Co₃O₄‐X, X = Zn, Al, Mg, Mn, Ni, Ce, Cr, etc.) can further tailor the electronic and redox properties of Co₃O₄ [29, 30, 31, 32, 33].

Despite advances in the design of Co₃O₄‐based catalysts, disentangling the roles of individual elements and precisely identifying the redox‐active sites under realistic conditions have proven difficult, as subtle changes are often masked by the bulk Co₃O₄ response [11, 16, 24, 28, 34]. Operando transient methods, particularly modulated excitation (ME) with phase‐sensitive detection (PSD), have therefore emerged as powerful tool for gaining molecular‐level insights into the structural and electronic dynamics of complex catalytic systems that are otherwise inaccessible using conventional methods [35, 36].

Direct N₂O decomposition into harmless N₂ and O₂ serves as an ideal probe reaction to interrogate interfacial properties in Co₃O₄‐X‐supported noble metal catalysts. Its mechanism is governed by structural, electronic, and redox properties of active sites that can be modulated at the metal–Co₃O₄ interface [37, 38]. Beyond the mechanistic relevance, its chemistry has remarkable environmental importance: N₂O is a potent greenhouse gas with a global warming potential ca. 310 times greater than CO₂ [39], and its emission has been identified as a key challenge in the future of carbon neutral transportation and energy sectors [40, 41]. Notably, the introduction of N₂O among regulated pollutants in the Euro VII emissions standards [42] and its potential formation in next‐generation combustion engines [40, 43, 44], highlights both its environmental significance and the technical challenges of its abatement. Hence, the development of effective technologies for N₂O mitigation has become crucial [44, 45].

In this work, we designed a mixed‐oxide RhO _x /Co₃O₄‐CeO₂ catalyst, exploiting the complementary role of its components. Rh provides highly active sites for N─O bond activation [46, 47], while Co₃O₄‐CeO₂ offers a redox‐active structure with tunable properties of the Rh/oxide interface [46, 48, 49, 50, 51]. To disentangle these contributions, we systematically investigated Co₃O₄, Co₃O₄‐CeO₂, RhO _x /Co₃O₄, and RhO _x /Co₃O₄‐CeO₂ catalysts using complementary ex situ characterization, DFT modeling, kinetic analyses, and N₂O decomposition as the probe reaction. By coupling operando x‐ray absorption spectroscopy (XAS)—mass spectrometry (MS) with ME‐PSD, we identify a cooperative reaction mechanism between RhO _x sites and Co ions sites on the most active catalyst in the series, RhO _x /Co₃O₄‐CeO₂.

2. Results and Discussions

2.1. Structural Characterization and Local Geometry

All catalysts exhibit the crystal structure of the cubic Co₃O₄ spinel (PDF#74‐1656; Fd‐3m space group symmetry; labeled as Co in Figure 1a), and display type‐IV adsorption–desorption isotherms (Figure S3). BET surface areas range from 64.9 m² g⁻¹ to 133.2 m² g⁻¹ (Table S1), increasing upon Ce incorporation (Co₃O₄‐CeO₂, labeled as CoCe) and decreasing after Rh impregnation (Rh/Co₃O₄, Rh/Co; Rh/Co₃O₄‐CeO₂, Rh/CoCe). No distinct reflections attributable to Rh‐containing phases are observed as indication of the high dispersion of Rh. Also, no CeO₂ phase is detected, but the characteristic peaks of Co₃O₄ broaden due to the decrease in crystallite size and increase in density of defect sites upon addition of Ce [52, 53]. Along this line, experimental pair distribution function (PDF; Figure 1b) confirms that the short‐atomic correlations correspond to the crystal structure of Co₃O₄ [54] (Figure S4) and provides direct evidence that Ce addition decreases the coherent domain size in CoCe and Rh/CoCe_, due to the presence of a more defective, distorted lattice [55, 56]. Rh and Ce also have no significant effect on the Raman spectra (Figure 1c), which display the characteristic phonon modes of the Co₃O₄ spinel [57], dominated by the vibration of Co³⁺‐O units in octahedral coordination (Co_Oh; A_1g, 690 cm⁻¹). Only this A_1g band redshifts and loses symmetry in Ce‐containing samples, consistent with the structural distortions in the Co₃O₄ lattice supported by powder x‐ray diffraction (PXRD) and PDF analysis. The observed changes reflect a modification of the local bonding environment of Co_Oh sites induced by CeO _x , resulting in more strained and oxidized Co sites [50]. Incorporation of Ce into the Co₃O₄ or formation of a mixed solution is unlikely due to the marked difference in ionic radii between Co³⁺ (55–61 pm) and Ce⁴⁺ (101–114 pm) [50, 58].

FIGURE 1.

(a) PXRD patterns, (b) pair distribution function data (PDF), and (c) Raman spectra of Co, CoCe, Rh/Co, and Rh/CoCe. (d and e) HAADF‐STEM images of Rh/CoCe_, and (f) EDX maps for Rh, Ce, and Co. The white circles in (e) highlight regions of higher contrast.

High‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) images of Rh/CoCe (Figure 1d) show that the material is composed of well‐crystallized Co₃O₄ host‐domains. The lattice spacing of ca. 0.241 nm in the HAADF‐STEM images and the corresponding inverse fast Fourier‐transform (IFFT) is assigned to the (311) planes of Co₃O₄. Instead of distinct RhO _x or CeO₂ nanoparticles, the isolated bright spots (high Z contrast) that are often aligned along the (311) lattice planes (Figure 1e) correspond to individual Rh or Ce atoms or very small clusters anchored to the surface of the spinel. Energy‐dispersive x‐ray (EDX) elemental maps (Figures 1f and S5) further confirm the high dispersion of Rh and Ce throughout the support. Together, the data confirms high dispersion and the absence of crystalline RhO _x or CeO _x domains. The short‐range nature of RhO _x and Rh⁺ species anchored on Co₃O₄ is further corroborated by the infrared spectra of adsorbed CO (Figure S6).

Qualitative analysis of the edge position and pre‐edge features of Co K‐edge x‐ray absorption near edge spectra (XANES, Figure 2a) measured at ambient conditions indicates a mixed‐valence state of Co between +2 and +3. Both Ce and Rh shift the average valence state of Co toward a more oxidized character compared to Co₃O₄, in the order Rh/CoCe ∼ CoCe > Rh/Co > Co, that is confirmed by the increased white‐line intensity. The intensity of the pre‐edge feature, which characterizes symmetry effects in the 1s→3d transition and is sensitive to local coordination symmetry, also decreases for the Ce‐ and Rh‐containing samples relative to Co₃O₄, suggesting increased centrosymmetry and reduced 3d–4p mixing in the local Co coordination environment [59]. Three main coordination shells in the Co K‐edge Fourier‐transformed (FT) extended x‐ray absorption fine structure (EXAFS) spectra of all samples (Figure 2b) are assigned to Co─O bonds (R = 1.55 Å), Co_Oh–Co_Oh pairs (2.57 Å), and Co_Td–Co_Td pairs (3.1 Å), consistent with a mixed‐valence Co₃O₄ spinel structure [32, 50]. Rh and especially Ce cause a slightly higher Co–O coordination number (N = 5.4–6) compared to Co₃O₄ (5.2; Figures S7 and S8; Table S2), indicating a more fully developed octahedral Co environment, in line with the pre‐edge observations. This is consistent with a higher average Co oxidation state and stronger Co–O covalency, likely due to Co charge re‐distribution following electron transfer from Co₃O₄ to Ce [50, 60]. Therefore, while the average structure of the spinel lattice is preserved, Ce and Rh induce local structural distortion in the Co₃O₄ lattice [50], in agreement with the above structural characterization (Figure 1). The similar Rh‐K edge XANES spectra of Rh/Co and Rh/CoCe (Figure 2c) display an absorption edge at higher energies than that of Rh₂O₃, revealing an average oxidation state higher than +3 [47, 61]. Furthermore, the higher oxidation state in Rh/CoCe than in Rh/Co indicates that Ce can act as an electron reservoir facilitating charge transfer between Rh and Co sites [62], leading to a higher oxidation state of Rh and Co. The corresponding Rh K‐edge FT‐EXAFS spectra (Figure 2d) are dominated by a first coordination shell of oxygen neighbors (phase‐uncorrected R = 1.91 Å). Both Rh‐impregnated samples show comparable Rh‐O shell intensities with coordination numbers of ca. 5.4 ± 0.4 (Figures S9 and S10; Table S3). A second coordination shell emerges at higher R values (phase‐uncorrected R = 2.4 Å) that is distinct from the Rh–Rh scattering of metallic Rh, suggesting the presence of Rh–Co and Rh–Ce interactions. Wavelet‐transform (WT)‐EXAFS analysis of Rh/Co and Rh/CoCe (Figure 2e) confirms this hypothesis and reveals two distinct intensity regions. The first corresponds to a Rh–O coordination shell with a local maximum at k = 7.74 Å⁻¹ and R = 1.56 Å, aligning with Rh₂O₃ and pointing toward the oxidized state of Rh. A second domain at k = 8.51 Å⁻¹ and R = 2.49 Å disagrees with the characteristic Rh–Rh scattering of metallic Rh (k = 10.37 Å⁻¹ and R = 2.37 Å), suggesting a second‐shell contribution from backscattering by lighter neighboring atoms [19, 24, 59]. In the case of Rh/CoCe, the latter maximum shifts to k = 8.06 Å⁻¹ and R = 2.49 Å, indicative of an enhanced contribution of Rh‐Co(Ce) interactions, likely arising from stronger interface interactions and a higher degree of Rh dispersion. Because Rh–O–Ce neighbors are expected at values of R = 3.64 Å and at higher k, the observed results are more consistent with a dominant Rh–O–Co contribution [63, 64]. Notably, EXAFS fitting (Figures S9 and S10; Table S3) is well described by explicit Rh–Co and Rh–O–Co scattering paths, further supporting short‐range RhO _x species rather than pure Rh–Rh contributions. The effect of Ce is reflected in a slightly longer Rh–Co distance (2.906 Å for Rh/Co vs. 2.920 Å for Rh/CoCe), consistent with the presence of a Rh–O–Co–O–Ce interface.

FIGURE 2.

(a) Normalized Co K‐edge XANES spectra and (b) corresponding phase‐uncorrected k ³‐weighted magnitude of the FT‐EXAFS spectra of Co, CoCe, Rh/Co, Rh/CoCe, and reference samples (LaCoO₃ and CoAl₂O₄). (c) Normalized Rh K‐edge XANES and (d) k ³‐weighted magnitude of FT‐EXAFS spectra of Rh/Co, Rh/CoCe, and reference spectra (Rh foil, scaled five times (/5), and Rh₂O₃), and (e) their corresponding k ³‐weighted wavelet‐transforms (WT).

2.2. Electronic Structure and Reducibility Properties

The decomposition of N₂O is considered to proceed via a redox mechanism, in which the desorption of surface oxygen (O₂) is critical. Therefore, catalyst reducibility and the nature of surface oxygen species are key parameters to rationalize the catalyst behavior.

The two distinct reduction events in the H₂ temperature‐programmed reduction (H₂‐TPR) thermogram of Co₃O₄ (Figure 3a) characterize the reduction of Co³⁺ to Co²⁺ (Co₃O₄ → CoO, ca. 265°C) and the subsequent reduction of Co²⁺ to metallic Co (ca. 355°C) [65]. The shift of the former peak to lower temperature (Co₃O₄ → CoO, ca. 257°C) in CoCe [51, 65, 66] is associated with redox interactions between Co³⁺/Co²⁺ and Ce⁴⁺/Ce³⁺ redox couples [56, 66, 67], and is accompanied by a broadening of the high‐temperature reduction peak, suggesting decreased reducibility of Co²⁺ [65], but also a possible contribution from the reduction of Ce⁴⁺ that is facilitated by H₂ spillover from Co. These findings are consistent with the Raman spectra [68, 69, 70] (Figure 1c), indicating that Ce primarily affects the Co³⁺–O environment (Co_Oh coordination) [71]. To confirm this, the incorporation of Rh in bulk Co₃O₄ was investigated by comparing the substitution of octahedral and tetrahedral sites (Figure S11). DFT simulations showed that the octahedral coordination of Rh is favored by 1.87 eV relative to the tetrahedral environment. This yields an oxidation state (+3.1) comparable to the XAS results, in contrast to that of the Rh in tetrahedral coordination (+2.7). Nonetheless, in line with previous studies [29, 32], the simulations also demonstrate that the incorporation of octahedral Rh induces a modification of the Co²⁺–O environment, specifically a slight contraction (average of 0.015 Å) of the surrounding tetrahedral Co²⁺–O bonds. However, this effect is expected to be limited, in agreement with the previous characterization results.

FIGURE 3.

(a) H₂ temperature programmed reduction (H₂‐TPR) of Co, CoCe, Rh/Co, and Rh/CoCe. (b) O 1s XPS spectra of Co, CoCe, Rh/Co, and Rh/CoCe and (c) Rh 3d XPS spectra of Rh/Co and Rh/CoCe.

The Rh‐containing catalysts (Rh/Co and Rh/CoCe) exhibit an additional low‐temperature feature due to reduction of RhO _x species [72, 73]. This peak appears at a lower temperature for Rh/CoCe than for Rh/Co, suggesting further improved reducibility due to the interaction between Rh, Co, and Ce. In both catalysts, the characteristic Co₃O₄ reduction peaks (Co³⁺ → Co²⁺ → Co⁰) collapse into a broader, single feature that shifts to lower temperature [74, 75]. This indicates enhanced oxygen mobility within the Co₃O₄ host lattice, as well as a promotional effect of Rh on the redox properties of Co (and Ce) species, likely due to Rh–Co(Ce) interactions [76, 77]. H₂ spillover from Rh to Co₃O₄ may also promote the migration of bulk oxygen to the surface [78].

The low temperature region of the O₂ temperature‐programmed desorption profiles (<400°C, O₂‐TPD, Figure S12) corresponds to the desorption of surface oxygen species (O_ads, i.e. O²⁻ and O⁻), while high‐temperature desorption (>400–700°C) is associated with the release of lattice oxygen (O_latt) [79, 80]. Focusing on O_ads, the incorporation of both Ce and especially Rh lowers the onset temperature for oxygen desorption. This implies a readier supply of active oxygen species at the surface, which is favorable for N₂O decomposition, as it reflects the ability to recombine and remove O atoms formed during reaction [46]. Consistently, the oxygen‐desorption capacity obtained from integration of the thermograms follows the same order as the reducibility observed in the H₂‐TPR profiles: Rh/CoCe > Rh/Co > CoCe > Co.

The oxidation state and oxygen lability were probed by x‐ray photoelectron spectroscopy (XPS, Figure 3b,c). The Co 2p spectra (Figure S13 and Table S4) display the characteristic spinel doublet consisting of components assigned to Co³⁺ (ca. 779.2 eV) and Co²⁺ (ca. 780.8–781.9 eV) and their satellites at ca. 10 and 6 eV from the main line, respectively [81]. The weak Co²⁺ satellite supports the predominant presence of a spinel Co₃O₄ phase [81, 82], consistent with the structural characterization. The surface Co²⁺/Co³⁺ ratio follows the order: Co (1.10) > Rh/Co (1.00) > CoCe (0.81) > Rh/CoCe (0.75), mirroring that obtained from Co K‐edge XANES, H₂‐TPR, and O₂‐TPD. This suggests that Rh and Ce stabilize Co³⁺ at the surface, enhancing oxygen mobility and the ability of the material to participate in Mars‐van Krevelen‐type reactions, involving lattice oxygen [33, 68, 83]. Three components were used to fit the O 1s spectra (Figure 3c and Table S5): lattice oxygen (O_I, 529.5 eV), surface chemisorbed oxygen (O_II, 530.5 eV), and hydroxyls (O_III, 533.0 eV) [68, 83]. The O_I component of Co₃O₄ (529.41 eV) shifts to lower energy upon addition of Rh (529.31 eV in Rh/Co) and Ce (529.33 eV in CoCe), while Rh/CoCe possesses the lowest O_I value (529.19 eV). Overall, these changes indicate the formation of more reactive and covalent metal–oxygen bonds [66], corroborating the above characterization.

The influence of Ce on the surface redox behavior is evident. In the Ce 3d region of both CoCe and Rh/CoCe, the fit of the signals using a combination of Ce³⁺ and Ce⁴⁺ components reveals partially reduced CeO _x (Figure S14 and Table S6) [84]. The spectra resemble those characteristic of ceria (CeO₂), clearly differing from the signature of atomically dispersed Ce [85], in agreement with the observation from XRD and PDF that Ce does not incorporate in the spinel lattice, but forms well‐dispersed CeO _x agglomerates visible by HAADF‐STEM but not large enough to be captured by XRD. In the presence of Rh, the Ce³⁺ fraction increases moderately from 0.36 (CoCe) to 0.40 (Rh/CoCe), pointing to an electronic coupling between Rh and the CeO₂ agglomerates. The magnitude of this change argues against a structure in which Rh is hosted on these CeO₂ domains [63]. Based on the structural characterization, the observed change is more consistent with a short‐range interaction between Rh and neighboring CeO₂, known for its dynamic redox behavior and ability to act as electron reservoir [86], with both phases mutually influencing their respective redox properties. The Rh 3d spectra can be fit with only Rh³⁺ (ca. 309 eV) and Rh⁴⁺ (ca. 310 eV) components [87, 88], with Rh³⁺/Rh⁴⁺ ratios (Rh/Co, 9.59; Rh/CoCe, 1.37) demonstrating that Ce increases the Rh⁴⁺ content (Table S4) and the electronic coupling between Rh and CeO₂.

DFT calculations were performed to gain atomic‐level insights into the experimentally observed redox behavior and to develop a rationale for the progressive reorganization of the electronic structure of Co₃O₄ upon addition of Ce and Rh, which ultimately impacts the surface and catalytic properties. Based on experimental characterization, CeO _x agglomerates on the Co₃O₄ surface were modeled as Ce₄O₇ clusters, while oxidized Rh was represented as atomically dispersed on the spinel host. The relative position of RhO _x and CeO _x was determined by systematically screening different configurations (Figure S1), with the most stable arrangement forming Rh─O─Co─O─Ce bonds. This configuration confirms the hypothesis of short‐range interfacial interactions between RhO _x and CeO _x , providing a structural basis to interpret the electronic coupling observed experimentally. Furthermore, the excellent agreement between the fitted EXAFS results (Figures S7–S10) and the calculated scattering paths from the DFT structures indicates that the proposed models are consistent with the local structures proposed based on the experimental data. The projected density of states (PDOS) of the surfaces was analyzed, focusing on the band centers of the different elements (O 2p, Co 3d, Rh 4d, and Ce 4f), which are considered descriptors of the oxygen lability and redox flexibility in metal oxides [89, 90]. The PDOS data are shown in Figure 4a, and the corresponding structural models in Figure 4b.

FIGURE 4.

Electronic structure analysis. (a) Projected density of states (PDOS) and (b) structural models of Co, CoCe, Rh/Co, and Rh/CoCe. (c) Relation between the oxidation state of Rh from the Bader charges (DFT) and XPS data of Figure 3. (d) Correlation between the centroid of the O 2p band (ε _O2p) from DFT and the lattice oxygen signal from the O 1s XPS narrow scans. The atoms in the DFT structures are identified in the legend.

The electronic structure of Co₃O₄ (Figure 4a) is dominated by Co─O─Co bonds. In a pristine spinel framework, the O 2p band lies well below the Fermi level (E_F) (ε _2p = −4.03 eV), and the Co 3d states split into a bonding (ε _3d = −3.21 eV) and an upper band (ε _3d* = +2.02 eV). This electronic configuration matches a partially covalent Co–O interaction and a moderate lattice oxygen mobility [89, 90], mirroring the structural characterization. In CoCe, interfacial Co─O─Ce bonds form with localized unoccupied Ce 4f manifolds (ε _4f* = +1.09 eV) located between the occupied O 2p (ε _2p = −3.87 eV) and the unoccupied Co 3d (ε _3d* = +2.01 eV) states. These Ce 4f states act as electron buffer, providing intermediate electron‐accepting reservoirs that reduce the effective metal‐oxygen gap [90]. The result is enhanced activation of metal─oxygen bonds, supported by the XPS and H₂‐TPR/O₂‐TPD data.

When RhO _x anchors to the surface of Co₃O₄, Rh–O–Co bridges are formed, leading to the reorganization of the electronic structure around the Rh states. In proximity of E _F, a high density of occupied Rh 4d states is observed compared to Co 3d (ε _3d = −3.29 eV), implying that less energy is required for RhO _x to participate in electron transfer [68]. The O 2p band shifts upwards (ε _2p = −3.71 eV), and the low‐lying Rh 4d antibonding states (ε _4d* = +1.15 eV) effectively become the lowest unoccupied levels (Co 3d ε _3d* = +2.20 eV). This decreases the effective metal–oxygen gap and improves the redox flexibility and mobility of oxygen [89, 90].

The most substantial restructuring of the electronic structure occurs when RhO _x and CeO _x coexist, and Rh–O–Co–O–Ce motifs are formed. The density of occupied Rh 4d states close to E _F increases with respect to Rh/Co, and the O 2p (ε _2p = −3.71 eV) and Co 3d (ε _3d = −3.02 eV) levels move upwards. The onset of the antibonding Rh 4d states located right above E_F lowers the upper‐band centroid (ε_4d* = +1.05 eV) and decreases the effective metal‐oxygen gap further. The Ce 4f states appear at higher energy (ε_{4f *} = +1.52 eV), followed by the Co 3d (ε_{3d *} = +2.26 eV). The proximity of the Ce 4f and Rh 4d states, mediated by bridging oxygen atoms, effectively couples Ce to Rh, in agreement with the increased content of Ce³⁺ sites observed from XPS. In a similar manner, the upward displacement of the Ce 4f* level toward the Co 3d* states strengthens the Ce–Co antibonding overlap, with the interconnecting oxygen atoms enabling efficient redox communication. Altogether, Rh 4d unoccupied states are now the lowest‐energy electron reservoir available for metal‐oxygen transfer, and Ce acts as a mediator of short‐range hybridization between Co and Rh [90]. This produces a single, integrated network in which electron redistribution between Rh, Ce and Co is energetically favored, stabilizing Rh in a higher, more reactive, oxidation state. The increase of the average Rh oxidation state from +3.09 (Rh/Co) to +3.42 (Rh/CoCe) observed in XPS is consistent with the values obtained from the DFT Bader charges of +3.11 (Rh/Co) and +3.27 (Rh/CoCe) (Figures 4c and S15).

The different bonding patterns and the resulting electronic coupling greatly impact the activation of metal–oxygen bonds and the reactivity of oxygen, which is experimentally probed by O₂‐TPD and H₂‐TPR. The O 2p states, representing the valence electrons most involved in metal‐oxygen bonding, shift upward from Co (−4.04 eV) to Rh/CoCe (−3.72 eV), indicating more reactive and delocalized electrons, correlating with a smaller energy gap with the metal bands and stronger hybridization [86]. These more delocalized and energetic electrons can screen the photoelectric core‐hole generated in the O 1s orbital more effectively, lowering the resulting binding energy [91], as reflected by the linear correlation observed between the DFT ε _2p and the XPS binding energies of the O 1s main photoemission line (Figure 4d). DFT thus demonstrates that Rh and Ce generate a synergistic mixed framework in which bond‐specific interactions induce cooperative electronic effects. The progression from Co–O to Co–O–Ce and to Co–O–Rh–O–Ce represents a systematic expansion of the metal–oxygen connectivity, defining the enhanced redox flexibility and lattice oxygen activation of Rh/CoCe.

2.3. N₂O Decomposition and Reaction Kinetics

We next investigate the effects of the formation of the Rh–O–Co–(O–Ce) interface on Co₃O₄ and establish structure–activity relationships for the N₂O decomposition (Figure 5a). Co₃O₄ exhibits moderate activity, with N₂O conversion commencing at ca. 275°C and a light‐off temperature (T ₅₀) of ca. 412°C. CoCe shows significantly earlier onset temperature at ca. 250°C, and a value of T ₅₀ of ca. 339°C, indicating improved redox behavior due to the increased oxygen mobility. Rh/Co further decreases the onset temperature (ca. 200°C) with a T ₅₀ value of ca. 312°C, and the highest activity is observed for Rh/CoCe, which reaches complete N₂O conversion in the 325°C–525°C temperature window, with an onset temperature of ca. 175°C and a T ₅₀ value of ca. 250°C. The order of reactivity suggested by the light‐off curves is corroborated by reaction rate values measured under differential reaction conditions (χ_N2O < 30%; Figure 5b and Table S7), allowing comparison among the samples in the absence of mass transport limitations. Relative to the Co₃O₄ spinel (0.67 mmol N₂O g_cat ⁻¹ h⁻¹), CoCe exhibits a 3.5‐fold activity enhancement (2.38 mmol N₂O g_cat ⁻¹ h⁻¹), Rh impregnation delivers a 4.3‐fold increase (3.01 mmol N₂O g_cat ⁻¹ h⁻¹), and Rh/CoCe shows a ca. 20‐fold increase (12.96 mmol N₂O g_cat ⁻¹ h⁻¹). These values align with, and in some cases surpass, previous data (Table S8), underscoring the critical role of Rh and CeO₂ in promoting N₂O decomposition. The pronounced variations in activity correlate with the differences in redox and structural properties among the catalysts, including enhanced lattice oxygen activation and covalency.

FIGURE 5.

Catalytic behavior of Co, CoCe, Rh/Co, and Rh/CoCe. (a) N₂O conversion as a function of reaction temperature. (b) Reaction rates measured under differential reactor conditions at 375°C. (c) Arrhenius plots derived from N₂O light‐off curves and (d) effect of N₂O partial pressure on reaction rates. Unless otherwise stated, all experiments were conducted with 500 ppmv N₂O, 2 vol% O₂, N₂ balance, at 1 atm, total flow rate of 100 mL min⁻¹, and variable catalyst mass (0–50 mg). E _a,app is expressed in kJ mol⁻¹.

This does not seem to be able to counteract the typical inhibitory effect of water and oxygen (Figure S16). H₂O (10 vol%) suppresses Rh/CoCe activity, shifting T ₅₀ from 250°C to 430°C (Figure S16a), while O₂ (10 vol%) lowers N₂O conversion by ca. 33% at 250°C (Figure S16b). Consistent with the O₂ inhibitory effect, Rh/CoCe loses also ca. 30% of N₂O conversion after 78 h on stream (Figure S17), likely due to accumulation of oxygen containing surface species [90, 92].

Arrhenius plots (Figure 5c) show consistently higher rates for Rh/CoCe across the examined temperature range. The apparent pre‐exponential factor (ln A _pp; Table S7) increases from 14.1 (Co₃O₄) to 15.6 (CoCe), consistent with structural differences influencing the effective collision frequency and/or exposed site density for N₂O at active Co sites. In contrast, the similar ln A _pp values for Rh/Co and Rh/CoCe (14.9) suggest a comparable Rh speciation/ensemble. Apparent activation energies (E _a,app) for the Co₃O₄ spinel (78.2 kJ mol⁻¹) align with reported values (Table S9) [65, 68, 93]. CoCe and Co₃O₄ show comparable E _a,app values (ca. 78 kJ mol⁻¹), suggesting that the same family of active sites is involved in N₂O decomposition over these materials and that the rate‐determining step (RDS) is not affected by CeO₂. In contrast, Rh in Rh/Co induces a significant decrease of E _a,app to ca. 69.5 kJ mol⁻¹; we associate this change with a modification in the nature of the active sites and the mechanistic pathway. This value decreases further to 64.1 kJ mol⁻¹ for Rh/CoCe, pointing toward a more efficient reaction pathway following the interaction between Rh and CeO₂.

Because N₂O decomposition operates at elevated temperatures, entropic effects are non‐negligible [48, 94]. Thus, we complemented the Arrhenius analysis with transition‐state theory using Eyring plots (Figure S18). For all catalysts, the enthalpy of reaction (∆H ^‡) values agree with the apparent Arrhenius E _a,app in the studied temperature range, consistent with a single dominant RDS. Rh/CoCe and Rh/Co exhibit essentially identical ∆H ^‡ (ca. 63 kJ mol⁻¹) and apparent ∆S ^‡ values (ca. −135 J mol⁻¹ K⁻¹), indicating similar per‐site kinetics; the higher overall rate therefore arises from greater accessible site density and/or coverage effects related to the Rh–O–Co–O–Ce interfacial effects discussed earlier, rather than from a reduced enthalpic barrier. By contrast, Co and CoCe display comparable ∆H ^‡ values (73 kJ mol⁻¹) but measurably different apparent ∆S ^‡ (−142 vs. −128 J mol⁻¹ K⁻¹). With the mechanism/RDS unchanged, the less negative apparent ∆S ^‡ of CoCe indicates a less constrained transition‐state ensemble, consistent with the improved lattice oxygen activation, accounting for the activity difference of Figure 5a,b.

To gain insights into the mechanism of N₂O decomposition, we evaluated the apparent reaction orders by examining the influence of the N₂O partial pressure on the reaction rates under kinetic control (Figure 5d). Typically, N₂O decomposition exhibits first‐order behavior in N₂O when the RDS involves N₂O dissociation. Deviations from first‐order kinetics are generally attributed to the influence of subsequent surface oxygen recombination [65, 95, 96]. Both Co and CoCe show apparent reaction order values of ca. 0.9, consistent with N₂O dissociation as the RDS, with minimal influence from the subsequent oxygen evolution step. Therefore, as observed from the Arrhenius analysis, Ce addition does not alter significantly the fundamental reaction pathway but modulates the structural and electronic properties of Co₃O₄ by improving reducibility and weakening Co─O bonds. Conversely, Rh reduces the apparent reaction order to ca. 0.74 for both Rh/Co and Rh/CoCe, in agreement with Rh‐based catalysts [47, 61, 88]. The deviation from first‐order kinetics suggests a mechanistic shift in the nature of the active sites and/or the RDS. While Rh is known to facilitate N‐O dissociation, it can also promote surface–oxygen accumulation, hindering O₂ desorption and rendering the recombination of surface oxygen species the new RDS. This variation highlights the critical role of Rh in altering the reaction pathway. Overall, the results are consistent with the Arrhenius parameters, confirming that Rh dictates the reaction pathway, whereas CeO₂ primarily influences the structural and electronic environment of the active sites. In summary, Rh and Ce exist as highly dispersed, short‐range species anchored to the Co₃O₄ host, with Rh in an oxidized (RhO _x ) state. Both components modulate the oxidation state and local coordination of Co via defects and strong electronic interactions at the Rh–O–Co–O–Ce interface.

2.4. Operando X‐Ray Absorption Spectroscopy (XAS) and Reaction Mechanism

Having established the structural and electronic properties of the most active catalyst in the series, we investigated the behavior of Rh/CoCe during N₂O decomposition using operando modulated excitation XAS experiments (ME‐XAS). Alternated cycles of N₂O/He (8000 ppm) and He mimic the redox nature of the N₂O decomposition reaction, comprising an oxidation‐half‐cycle (OHC) and a reduction‐half‐cycle (RHC). Because the apparent N₂O reaction order for Rh/CoCe indicates that Rh is responsible for N₂O activation and decomposition, the interaction of N₂O with the highly dispersed RhO _x species was probed by Rh K‐edge XANES. The average time‐resolved Rh K‐edge XANES spectra (Figure 6a, top) show a subtle positive shift of the absorption edge under N₂O‐rich conditions, consistent with oxidation of RhO _x (RhO _x → Rh ^{n +1}O _{x +1}) and a shift to lower energy during the He half‐cycle, indicating reduction of Rh ^{n +1}O _{x +1} to RhO _x . Phase‐sensitive detection (PSD; Figure 6a, bottom) enhances the sensitivity to these subtle changes providing unambiguous evidence of a reversible redox process (Figure S19), wherein N₂O induces the oxidation of Rh (2N₂O → 2N₂ + 2O; OHC), which is marked by a positive feature at ca. 23231 eV, followed by the recombination/desorption of O₂ that drives auto‐reduction of Rh (2O → O₂; RHC) [47, 61, 88], in agreement with the rapid N₂ evolution (m/z = 28) in the OHC and a slower O₂ release (m/z = 32) in the RHC (Figures S20 and S21). This behavior further points toward the RHC as RDS. Similar results were obtained for Rh/Co_, albeit with lower PSD amplitude (Figure S22), validating the Rh‐mediated redox cycle consistent with the similar kinetic fingerprints of Rh/Co and Rh/CoCe (Figure 5c,d).

FIGURE 6.

Modulated excitation experiments with alternate N₂O/He and He cycles over Rh/CoCe. (a) Time‐resolved Rh K‐edge XANES spectra (top) and phase‐resolved spectra (bottom). (b) Time‐resolved Co K‐edge XANES spectra (top) and phase‐resolved spectra (bottom). The dashed line is the scaled difference between the spectra of Co₃O₄ and CoAl₂O₄ references. (c) Rh K‐edge phase‐uncorrected k ²‐weighted magnitude of the FT‐EXAFS spectra of Rh/CoCe in N₂O/He and He environments. (d) Co K‐edge phase‐uncorrected k ²‐weighted FT‐EXAFS spectra in N₂O/He and He environments. Conditions: 8000 ppm N₂O, balance He, 1 atm, total flow rate of 100 mL min⁻¹, catalyst mass of 25 mg.

The intensity of the signal emphasized by PSD at 23231 eV scales linearly with N₂O partial pressure (p _N2O; n _N2O = 0.37; Figures S23 and S24), consistent with the measured reaction orders (n _N2O < 0.9) and the rate‐determining nature of the RHC, reflecting the inhibitory effect of surface O‐species accumulation at increasing p _N2O values [96]. Complete suppression of the PSD response in an identical ME‐PSD experiment conducted in excess of O₂ in the feed (Figure S25) further verifies that, on Rh‐based catalysts, the RHC is dominated by inhibition of the Rh redox cycle due to oxygen build‐up, in line with the catalytic experiments (Figure S16b) [74, 97].

In identical ME‐XAS experiments at the Co K‐edge (Figure 6b, top), N₂O induces a shift of the absorption edge toward lower energies, thus reduction of Co species, while in He the edge shifts to higher energies, indicating reoxidation. The PSD data clearly demonstrates the reversible nature of this process (Figure 6b, bottom) with spectral features closely resembling the spectral difference between Co₃O₄ and CoAl₂O₄. Therefore, Co species cycle between Co³⁺ (Co_Oh) and Co²⁺ (Co_Td) environments during N₂O decomposition on Rh/CoCe.

Remarkably, the response at the Co K‐edge made visible by PSD is shifted by about 130° relative to that at the Rh K‐edge (Figure S26) providing direct operando evidence of a coupled Rh–O–Co redox mechanism, wherein RhO _x and CoO _x species undergo sequential oxidation–reduction events, beginning with oxidation of RhO _x followed by reduction of CoO _x . This cooperative mechanism is consistent with the interfacial effects inferred from the structural, electronic, and surface characterization results.

Variations in the local environment during N₂O decomposition were probed by EXAFS. As shown in Figure 6c, the magnitude of the k ²‐weighted Rh K‐edge spectra of Rh/CoCe reveals that upon N₂O introduction, the Rh–O contribution slightly increases, consistent with the oxidation of RhO _x species in the N₂O‐rich period and a higher Rh–O coordination number, likely arising from oxygen deposition following N─O bond cleavage. More pronounced structural changes were observed in the magnitude of the k ²‐weighted Co K‐edge spectra (Figure 6d). The amplitude of the Co–O contribution decreases upon contact with N₂O, concomitant with an increase in the Co–Co_Td scattering contribution. The attenuation of the Co–O path agrees with the partial reduction of Co³⁺ _Oh sites indicated by XANES, together with increased disorder and/or a lowered average coordination associated with defect formation. The gain in Co‐Co_Td is ascribed to lattice rearrangement upon formation of oxygen vacancies (V_O) [60]. These results indicate that Co₃O₄ experiences dynamic and reversible structural changes that are complementary, though reciprocal, to those of Rh, underscoring their coupled redox behavior during N₂O decomposition.

Based on our kinetic analysis and operando ME‐XAS, and supported by previous studies [47, 74, 88, 98, 99], we propose a cooperative RhO _x /CoO _x reaction mechanism for N₂O decomposition over Rh/CoCe (Equations (1), (2), (3)). N₂O decomposition initiates with the OHC via dissociative adsorption of N₂O on RhO _x (Rh³⁺/Rh⁴⁺), leading to N─O bond cleavage, N₂ release, and oxidation of RhO _x (1). The RHC proceeds via auto‐reduction of oxygen‐rich RhO _x species (Rh ^{n +1}O _{x +1}) through a hetero‐recombination step, in which lattice oxygen from Co─O bonds in Co₃O₄ reacts with Rh‐bound oxygen to form O₂, alongside reduction of Rh and creation of an oxygen vacancy (V_O) (2). Finally, regeneration of the Co₃O₄ lattice through replenishment of V_O by N₂O produces the second N₂ molecule (3).

$$\mathrm{Rh}-{\mathrm{O}}_x+{\mathrm{N}}_2\mathrm{O}\to \mathrm{Rh}-{\mathrm{O}}_{x+1}+{\mathrm{N}}_2$$ (1)

$$\mathrm{Rh}-{\mathrm{O}}_{x+1}+{\mathrm{Co}}^{3+}-{\mathrm{O}}_{\mathrm{latt}}\to \mathrm{Rh}-{\mathrm{O}}_x+{\mathrm{Co}}^{2+}-{\mathrm{V}}_{\mathrm{O}}+{\mathrm{O}}_2$$ (2)

$$\mathrm{C}{\mathrm{o}}^{2+}-{\mathrm{V}}_{\mathrm{O}}+{\mathrm{N}}_2\mathrm{O}\to \mathrm{C}{\mathrm{o}}^{3+}-{\mathrm{O}}_{\mathrm{latt}}+{\mathrm{N}}_2$$ (3)

This cooperative Rh/Co sequence is consistent with a Mars‐van Krevelen mechanism, previously reported in the context of CO oxidation on Co₃O₄ and N₂O decomposition on CuO‐Co₃O₄ catalysts [100].

To further clarify the role of CeO _x , we used DFT to track charge redistribution along the key steps of the catalytic cycle on Rh/CoCe and, for comparison, on Rh/Co (Figure S27a,b). For both catalysts, Rh is oxidized upon N₂O adsorption and cleavage (Equation 1) and reduced during O₂ desorption (Equation 2), in agreement with the experimental evidence. However, the presence of CeO _x significantly modifies the electronic structure along the reaction sequence. On the clean slab, Rh is more oxidized in Rh/CoCe (+1.26 |e|) than in RhCo (+1.20 |e|), leading to a more favorable N₂O adsorption by 1.43 eV. Following N─O bond cleavage (Equation 1), N₂ evolves and the remaining oxygen atom migrates to the surface forming a Rh–O–O–Co peroxide intermediate involving the Co³⁺ _Oh sites of Co₃O₄ [68, 69, 70].

While this peroxide intermediate forms on both catalysts, it is more stabilized on Rh/CoCe, where the peroxo species binds to Rh through a single bond. Upon N₂O activation, the Ce³⁺/Ce⁴⁺ redox couple undergoes partial reduction (average Ce charge changes from +2.28 |e| to +2.18 |e|), with Rh and Co mediating charge transfer from CeO _x to the peroxo intermediate, resulting in more reduced oxygen species (average per O atom: −0.80 |e| and −0.37 |e| for Rh/CoCe and Rh/Co, respectively). Subsequently, after O₂ desorption, CeO _x moderates the reduction of interfacial Co³⁺ _Oh sites, enabling them to retain a higher oxidation state compared to Rh/Co (+1.34 |e| vs. +1.16 |e|). This stabilization of both the initial and the final state lowers the thermodynamic cost of O₂ release (Equation 2) by 0.71 eV compared to Rh/Co (Figure S27c). Taken together, these results show that, while Rh³⁺/Rh⁴⁺ sites drive N₂O activation, CeO _x promotes interfacial charge transfer via bridging oxygen atoms in the Co₃O₄ host, sustaining the redox cycle and enhancing catalytic activity.

3. Conclusion

The tunable structure and electronic properties of spinel metal oxides enable the design of a RhO _x /Co₃O₄‐CeO₂ catalyst with RhO _x and CeO _x species anchored and stabilized at short‐range on Co₃O₄. This engineered interface facilitates the formation of an expanded metal oxygen Rh–O–Co–O–Ce connectivity that enhances redox properties, promoting the weakening of the Co─O bonds and the activation of the lattice oxygen, delivering an efficient catalyst for the N₂O decomposition. Distinct reaction pathways for Rh‐containing catalysts compared to Co₃O₄ and Co₃O₄‐CeO₂ highlight the central role of Rh in the reaction mechanism in facilitating N‐O bond dissociation and modifying the rate‐determining step. Operando XAS measurements confirm a cooperative RhO _x ‐CoO _x Mars‐van Krevelen mechanism, in which the Rh redox cycle is sustained by electron/oxygen exchange with the Co₃O₄‐(CeO _x ) matrix, underscoring the role of operando techniques enhanced by advanced experimental protocols for uncovering surface processes and revealing mechanistic details in complex catalysts. This work emphasizes the importance of engineered metal oxide–metal oxide interfacial interactions in the design of effective catalysts for industrially relevant redox reactions and provides a platform for the development of improved oxygen‐based catalysts.

Author Contributions

Daniel C. Cano‐Blanco: conceptualization, investigation, writing – original draft, methodology, formal analysis, data curation, writing – review and editing. Silvio Bellomi: conceptualization, investigation, writing – original draft, methodology, formal analysis, writing – review and editing. Daniele Bonavia: formal analysis, investigation, writing – original draft, data curation, writing – review and editing. Ivo Alxneit: investigation, formal analysis, writing – review and editing. Kirill A. Lomachenko: investigation, writing – review and editing. Karin Föttinger: writing – review and editing, supervision, funding acquisition. Oliver Kröcher: writing – review and editing, supervision. Davide Ferri: supervision, project administration, writing – review and editing, funding acquisition, conceptualization.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information [99, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125]. Supporting File: anie73090‐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.