Redox-Driven Exsolution and Dissolution Behavior of High-Entropy Spinel Catalysts: A Comparative Study of MnFeCoNiCuO _x and MnCoNiCuZnO _x

Abstract

High-entropy oxides (HEOs) offer tunable redox chemistry and thermal stability for catalytic applications. Here, we compare two spinel-type HEOs, MnFeCoNiCuO _x and MnCoNiCuZnO _x , with similar configurational entropy but different redox behaviors under reverse water–gas shift (RWGS) conditions. Only MnFeCoNiCuO _x exhibits reversible exsolution and reincorporation of Fe/Co/Ni/Cu alloy nanoparticles (NPs) during H₂–CO₂ cycling, as confirmed by in situ X-ray absorption spectroscopy and wavelet-transformation. This dynamic restructuring correlates with higher concentrations of oxygen vacancy and exsolved Fe/Co/Ni/Cu alloy NPs, resulting in higher RWGS activity above 400 °C. In contrast, redox-inert Zn²⁺ in MnCoNiCuZnO _x suppresses lattice flexibility and alloy formation. These findings underscore that redox-active cations, rather than entropy alone, govern the regenerative behavior and catalytic performance in HEO systems.

1. Introduction

High-entropy oxides (HEOs) are single-phase crystalline materials comprising five or more cations in near-equimolar ratios, randomly distributed over crystallographic sites. − The resultant high configurational entropy thermodynamically stabilizes solid solutions, enabling the formation of otherwise metastable or unstable oxides. This entropy-induced stabilization also imparts structural tunability, distinctive defect chemistry, and enhanced thermal stabilityattributes that position HEOs as promising catalytic platforms for reactions such as the oxygen reduction and evolution reactions. ,

A key feature of HEO chemistry is the dynamic behavior of cations, enabling exsolution and redissolution in redox environments and resulting in self-regenerative structures. This phenomenon was observed in the early 2000s, when Daihatsu Motor and collaborators reported a self-regenerative catalystPd-doped LaFeCoO₃exhibiting reversible Pd nanoparticles (NPs) migration under oxidative and reductive conditions. , Similar behaviors were later reported with Rh- and Pt-doped perovskites, where B-site cations undergo exsolution and reincorporation. Other perovskite systems have shown comparable dynamics. For instance, the Bao group used in situ techniques such as X-ray diffraction (XRD) and scanning transmission electron microscopy to demonstrate reversible exsolution and redissolution of CoFe NPs in LaSrCoFeMoO₃. , Vert et al. studied the redox behavior of Ni in chromites, linking it to electrochemical performance in H₂/CH₄-fueled solid oxide fuel cells. Key factors enabling redissolution of exsolved NPs include substoichiometric A-site control, B-site cation selection, and tailored pre/post-treatment of perovskites. ,

Configurational entropythe diversity of molecular conformations and stacking in oxidesmay enable HEOs to function as self-regenerative catalysts. This is attributed to two effects: (1) structural disorder lowers Gibbs free energy, stabilizing the host lattice (support), and (2) entropy gain promotes the redissolution and oxidative reformation of NPs into their parent oxides. Zhao et al. and Hou et al. demonstrated this in the redox cycles of CO₂ hydrogenation: the former observed Cu/Co/Ni alloy exsolution and dissolution in Co₃MnNiCuZnO _x , the latter in Co/Fe/Cu/Ni alloys from Zr_0.5(NiFeCuMnCo)_0.5O _x . Shao et al. observed reversible cyclic evolution–dissolution of Ni, Fe, and Co to form Ni/Fe/Co alloys on halite-structured (MgCoNiMnFe)­O _x in dry reforming. Light-induced exsolution–dissolution of Co/Ni alloy in Rh-supported CoNiFeZnCrO _x in the solar-driven dry reforming of methane was revealed lately.

Despite increasing interest in redox-active HEOs, the mechanisms underlying alloy formation during reduction remain insufficiently understood. It is not yet established whether HEOs with comparable configurational entropy (ΔS _config.) necessarily exhibit similar exsolution behavior. These open questions motivate a closer examination of how cation chemistry, rather than entropy alone, dictates the structural dynamics and catalytic function under redox cycling.

In this study, we investigate the exsolution and dissolution behavior of exsolved elements in spinel HEOs, i.e., MnFeCoNiCuO _x and MnCoNiCuZnO _x , with close configurational entropies (ΔS _config. = 12.4 and 12.3 J/mol/K, respectively) under reducing (H₂) and oxidizing (CO₂) atmospheres. Both compositions are chosen based on their successful synthesis as single-phase, entropy-stabilized spinel HEOs to provide a structurally consistent basis for comparison. The redox-driven exsolution and redissolution behavior of HEOs is dependent on the cation chemistry, even when configurational entropy is similar. The MnFeCoNiCuO _x catalyst undergoes reversible exsolution of Fe/Co/Ni/Cu alloy domains under H₂, followed by reincorporation into the spinel lattice under CO₂. This dynamic behavior enhances the adsorption of CO₂ through the generation of oxygen vacancies and promotes hydrogenation via the exsolved alloy NPs, collectively leading to an increased reverse water–gas shift (RWGS) rate. In contrast, the redox-inactive Zn²⁺ in MnCoNiCuZnO _x limits structural regeneration, resulting in low RWGS activity. These findings establish that configurational entropy alone does not dictate redox dynamics or catalytic performance; instead, the rational incorporation of redox-active cations is key to unlocking the full potential of HEOs for regenerative CO₂ catalysis.

2. Experimental Section

2.1. Materials

Nickel chloride anhydrous (98%, Sigma-Aldrich), copper chloride anhydrous (98%, SHOWA), cobalt chloride hexahydrate (98%, Sigma-Aldrich), zinc chloride anhydrous (98%, Alfa Aesar), iron chloride hexahydrate (97%, Alfa Aesar), sodium hydroxide (97%, SHOWA), and potassium permanganate (99%, SHOWA) were used as received.

2.2. Catalyst Synthesis

The MnCoNiCuZnO _x and MnFeCoNiCuO _x catalysts were synthesized using a solid-state ball-milling method. Briefly, 6 mmol of CoCl₂·6H₂O, 2 mmol of NiCl₂, 2 mmol of CuCl₂, and 2 mmol of FeCl₃ (for MnFeCoNiCuO _x ) or 2 mmol of ZnCl₂ (for MnCoNiCuZnO _x ) were added to a 45 mL zirconium oxide milling jar together with zirconium oxide milling balls with a ball-to-material ratio of 10:1. The mixture was milled in a planetary ball mill (Fritsch Pulverisette 7) at 500 rpm for 2 h. Subsequently, 24 mmol (for MnFeCoNiCuO _x ) or 22 mmol (for MnCoNiCuZnO _x ) of sodium hydroxide was added, and the mixture was milled for another 2 h. Afterward, 2 mmol of KMnO₄ was added, followed by an additional 2 h of milling. The resulting solid was washed with deionized water, separated by centrifugation, dried overnight at 80 °C, and calcined at 600 °C for 2 h (5 °C/min) in an air stream (25 mL/min).

Some characterizations require both fresh and reduced samples, and a reduction step was conducted prior to the analysis. Each catalyst was reduced in a 20% H₂/N₂ stream from room temperature to 450 °C (10 °C/min) and held at 450 °C for 2 h in a fixed-bed system. The reduced HEOs were denoted as r-MnCoNiCuZnO _x and r-MnFeCoNiCuO _x .

2.3. Characterization Methods

Catalyst composition was analyzed via inductively coupled plasma atomic emission spectroscopy (ICP-AES, Kontron S-35), and textural properties were measured by using N₂ physisorption (Micromeritics ASAP 2020). Elemental distribution was examined by high-resolution transmission electron microscopy (JEOL JEM-2100F CS) equipped with an Oxford X-Max 20 EDS detector. Crystallinity was assessed by powder XRD (Rigaku SmartLab, Cu Kα). X-ray absorption spectra (XAS) at the Mn, Fe, Co, Ni, Cu, and Zn K-edges were collected in fluorescence mode at the TPS 32A beamline of the National Synchrotron Radiation Research Center (NSRRC). Data processing employed Athena (Demeter v0.9.26); wavelet transforms were applied to k ²-weighted XAS using a Morlet function in the k-range of 3–17 Å^–1. X-ray photoelectron spectroscopy (XPS) was conducted on a PHI 5000 VersaProbe spectrometer (Al Kα, 1486.6 eV). To avoid air exposure, samples were transferred via a sealed chamber for quasi-in situ analysis. Binding energies were calibrated to the C 1s peak at 284.8 eV.

Temperature-programmed reduction (H₂-TPR) and CO₂ desorption (CO₂-TPD) were performed by using an AutoChem II system (Micromeritics) with a thermal conductivity detector. Samples were prereduced at 450 °C for 2 h. After CO₂ saturation and He purging, CO₂-TPD was conducted under He flow (25 mL/min) at a 10 °C/min ramping rate while H₂-TPR was conducted in a 10% H₂/Ar stream (25 mL/min) at a 10 °C/min ramping rate.

Fourier-transform infrared spectra were recorded on a Nicolet iS50 instrument (Thermo Scientific). In situ DRIFTS experiments were performed using a Praying Mantis cell (Harrick Scientific) with CO₂–H₂ switching: samples were heated from 50 to 350 °C in CO₂ (20 mL/min), then switched to H₂ (20 mL/min) at 350 °C and held isothermally for 10 min.

In situ XAS was also conducted at the NSRRC (TPS 32A). Samples were sealed in a 1.5 mm capillary, heated using a heat gun, and analyzed at various temperatures under H₂ and CO₂ streams to monitor Fe and Zn K-edge evolution. The Fe K-edge spectra were collected at room temperature, 350 °C, 400 °C, and 450 °C. Zn K-edge was recorded only at room temperature and 450 °C under comparable conditions.

2.4. Activity Evaluation

Catalytic performance was evaluated in a fixed-bed reactor system. , Approximately 0.03 g of catalyst was loaded into quartz wool plugs. The reaction was carried out under a CO₂/H₂/N₂ mixture (12.5/37.5/50.0 v/v/v) at atmospheric pressure, within a temperature range of 350–500 °C. Each activity data point reflects the average of three independent measurements; the 95% confidence interval, calculated from these replicates, is shown as the error bar. A gas hourly space velocity (GHSV) of 40,000 mL/g_cat/h was applied, and the reactor effluent was kept at 150 °C to avoid condensation. CO₂ conversion and product selectivity (CO and CH₄) were calculated according to established equations. ,

3. Results

3.1. Unreduced MnFeCoNiCuO _x and MnCoNiCuZnO _x

Table S1 summarizes the elemental compositions determined by ICP-AES and the textural properties measured by N₂ physisorption. The molar ratios of MnFeCoNiCuO _x and MnCoNiCuZnO _x were approximately 1:1:3:1:1 and 1:3:1:1:1, respectively, in good agreement with the target stoichiometries. By resorting to the elemental compositions, the ideal mixing formula (ΔS _config. = −R∑ _i = 1 x _i ln x _i ) is used to calculate the changes of configurational entropy for MnFeCoNiCuO _x (12.4 J/mol/K) and MnCoNiCuZnO _x (12.3 J/mol/K). Both HEOs exhibited comparable surface areas (46.8 and 52.6 m²/g) and pore volumes (0.36 and 0.28 cm³/g), indicating similar porous characteristics (see Figure S1).

Figure a,b presents the TEM-EDS elemental mapping of MnFeCoNiCuO _x and MnCoNiCuZnO _x , respectively. Both samples exhibit uniform elemental dispersion, consistent with the formation of single-phase HEOs. The widespread Cu signals are partially attributed to the use of a Cu grid as the sample holder; however, the confinement of Cu signals within the particle boundary also suggests its incorporation into the HEO structure.

1.

TEM images and EDS elemental mapping results of (a) MnFeCoNiCuO _x and (b) MnCoNiCuZnO _x , (c) XAS K-edge spectra of Fe, Mn, Co, Ni, and Cu of MnFeCoNiCuO _x , (d) XAS K-edge spectra of Zn, Mn, Co, Ni, and Cu of MnCoNiCuZnO _x , (e) XPS spectra of Fe 2p, Mn 2p, Co 2p, Ni 2p, Cu 2p, and O 1s photolines of MnFeCoNiCuO _x , and (f) XPS spectra of Zn 2p, Mn 2p, Co 2p, Ni 2p, Cu 2p, and O 1s photolines of MnCoNiCuZnO _x .

Figure S2 shows the XRD patterns of MnFeCoNiCuO _x and MnCoNiCuZnO _x . Both catalysts had diffractions at 30.6°, 36.0°, 43.7°, 57.4°, 57.7°, and 63.5°, corresponding to the (220), (311), (400), (422), (511), and (440) planes of Fd3̅m (227) spinel Co₃O₄ (JCPDS #80-1540).

Figure c,d exhibits the XAS K-edge spectra of each cation of MnFeCoNiCuO _x and MnCoNiCuZnO _x . The normalized intensity of the pre-edge peak is used as an indicator of the extent of deviation from centrosymmetry: the higher the intensity, the higher the portion of cation located in the tetrahedral position. ,, The normalized pre-edge intensity of MnFeCoNiCuO _x showed a decreasing trend as Fe (0.195) > Mn (0.105) > Co (0.093) > Ni (0.025) > Cu (0); the pre-edge intensity of MnCoNiCuZnO _x : Mn (0.124) > Co (0.113) > Ni (0.023) > Cu (0). These trends suggest that both HEOs catalysts should be complex spinel structures with partial inversion. , For MnFeCoNiCuO _x , Fe cations coexist in tetrahedral and octahedral units, while the Franklinite-like Zn spectrum, which is similar to an ideal normal spinel structure, indicated that Zn cations were in tetrahedral units.

Figure e,f shows the XPS surface analysis of MnFeCoNiCuO _x and MnCoNiCuZnO _x and Table S2 presents the relative surface composition of cations. For MnFeCoNiCuO _x , the Fe 2p spectrum exhibited mixed Fe³⁺ (55%) and Fe²⁺ (45%). The Mn 2p spectrum revealed Mn⁴⁺ and Mn³⁺ at 14% and 86%, Co 2p showed Co³⁺ (54%) and Co²⁺ (46%), Ni appeared as Ni²⁺ (100%), and Cu appeared as Cu²⁺ (95%) and Cu⁰ (5%). For MnCoNiCuZnO _x , Zn 2p showed predominantly Zn²⁺. Mn⁴⁺ (13%) and Mn³⁺ (87%), Co³⁺ (48%) and Co²⁺ (52%), Ni²⁺ (100%), Cu²⁺ (90%), and Cu⁰ (10%) were observed. The O 1s spectra of both catalysts showed three components: lattice oxygen (O^α, 529.4 eV), oxygen in surface vacancies (O^β, 531.0 eV), and hydroxyl/surface-adsorbed oxygen (O^γ, 534.0 eV). , The O^β/O^α ratio was higher in MnFeCoNiCuO _x (1.1) than in MnCoNiCuZnO _x (0.8), indicating a greater concentration of the surface oxygen vacancy (O_v) of the former.

Figure a exhibits the CO₂-TPD profiles of MnFeCoNiCuO _x and MnCoNiCuZnO _x . Both catalysts had similar CO₂ desorption profiles, peaked at low- (ca. 125 °C) and medium- (ca. 180–190 °C) regions. The amounts of desorbed CO₂ at low- and medium-temperatures were 121.3 and 135.7 μmol/g for MnFeCoNiCuO _x and 135.8 and 143.6 μmol/g for MnCoNiCuZnO _x , respectively. MnCoNiCuZnO _x showed a slightly higher CO₂ desorption amount (279.4 μmol/g) than that of MnFeCoNiCuO _x (257.0 μmol/g).

2.

(a) CO₂-TPD profiles of MnFeCoNiCuO _x and MnCoNiCuZnO _x , (b) H₂-TPR profiles of MnFeCoNiCuO _x and MnCoNiCuZnO _x , and (c) CO₂-TPD profiles of r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x . Conditions: CO₂-TPD was recorded in a He stream (25 mL/min) at a ramp rate of 10 °C/min; H₂-TPR, 10% H₂/Ar (25 mL/min) at a ramp rate of 10 °C/min.

Figure b shows the H₂-TPR profiles of MnFeCoNiCuO _x and MnCoNiCuZnO _x . Both catalysts exhibit multiple reduction peaks: a minor shoulder at 178 °C (Cu²⁺ → Cu⁰), followed by overlapping peaks at 340–350 °C (Ni²⁺ → Ni⁰) and 380–410 °C (Co cations reduction). , An additional shoulder at 505 °C appears only for MnFeCoNiCuO _x , corresponding to Fe³⁺ reduction. Total H₂ consumption was 10.5 mmol/g for MnFeCoNiCuO _x and 9.1 mmol/g for MnCoNiCuZnO _x . The 1.4 mmol/g difference is attributed to the presence of reducible Fe³⁺ in the former, whereas Zn²⁺ in the latter seemed to be unreducible below 800 °C.

3.2. Reduced MnFeCoNiCuO _x and MnCoNiCuZnO _x

Figure a,b shows representative TEM images and EDS mappings of r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x NPs, respectively. Compared with the fresh forms, these reduced particles exhibit uneven elemental distributions, particularly in the highlighted red-boxed areas, consistent with localized exsolution sites. For r-MnFeCoNiCuO _x , O and Mn were nearly undetectable in the highlighted region, while for r-MnCoNiCuZnO _x , O, Mn, and Zn were nearly unseen. This implied that the reduction treatment induced the relocation of some elements. That is, during the reduction, Fe, Co, Ni, and Cu can be redispersed in r-MnFeCoNiCuO _x while Co, Ni, and Cu can be relocated in r-MnCoNiCuZnO _x .

3.

TEM images and EDS elemental mapping results of (a) r-MnFeCoNiCuO _x and (b) r-MnCoNiCuZnO _x , (c) XPS spectra of Fe 2p, Mn 2p, Co 2p, Ni 2p, Cu 2p, and O 1s photolines of r-MnFeCoNiCuO _x , and (d) XPS spectra of Zn 2p, Mn 2p, Co 2p, Ni 2p, Cu 2p, and O 1s photolines of r-MnCoNiCuZnO _x .

Figure S2 displays the XRD patterns of r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x . Both reduced catalysts exhibit a doublet near 44.0°, along with several minor diffraction peaks, suggesting the formation of Co/Ni/Cu-based alloy phases.

Figure c,d shows the XPS spectra of r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x and Table S2 summarizes the relative surface cation compositions. The spectrum of each element was like its unreduced counterpart, with a decreased ratio of high-to-low cation/metal oxidation state. For r-MnFeCoNiCuO _x , the Fe 2p spectrum exhibited Fe³⁺ (46%) and Fe²⁺ (54%). The Mn 2p spectrum showed Mn³⁺ (45%) and Mn²⁺ (55%), Co 2p showed Co³⁺ (28%), Co²⁺ (64%), and Co⁰ (8%), Ni showed Ni²⁺ (96%) and Ni⁰ (4%), and Cu showed Cu²⁺ (64%) and Cu⁰ (36%). For r-MnCoNiCuZnO _x , the Zn 2p spectrum was nearly identical to its unreduced form, showing a characteristic Zn²⁺ photoline. The other cations showed compositions of Mn³⁺ (34%) and Mn²⁺ (66%), Co³⁺ (41%), Co²⁺ (40%) and Co⁰ (19%), Ni²⁺ (97%) and Ni⁰ (3%), and Cu²⁺ (65%) and Cu⁰ (35%).

The O^β/O^α ratio of r-MnFeCoNiCuO _x (1.7) was higher than that of r-MnCoNiCuZnO _x (1.1), and both values were higher than those of the pristine forms. That is, the reduction of HEOs develops the corresponding O_v sites, and r-MnFeCoNiCuO _x showed a higher relative O_v composition than that of r-MnCoNiCuZnO _x .

Figure c shows the CO₂-TPD profiles of r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x . The low- and medium-temperature desorption peaks shifted from ca. 125 to 140–170 °C and from ca. 180–190 to 230–270 °C, respectively, compared to their unreduced counterparts (see Figure a). The total amount of desorbed CO₂ from the low- and medium-temperature desorption peaks increased from 257.0 to 392.1 μmol/g for r-MnFeCoNiCuO _x and from 279.4 to 288.1 μmol/g for r-MnCoNiCuZnO _x . This corresponds to an enhancement of 135.1 μmol/g in the CO₂ adsorption capacity for r-MnFeCoNiCuO _x , more than an order higher than the increase observed for r-MnCoNiCuZnO _x (8.7 μmol/g).

To further understand the redispersion/exsolution behaviors of r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x , the wavelet-transformed XAS (WT-XAS) analysis of their reduced forms was conducted, shown in Figure . The WT-XAS 2D contour plots of Co, Ni, Cu, and Fe foils and ZnO were included for comparison. The WT-XAS 2D contour plot exhibits not only the interatomic distance (R space) but also backscattering amplitude (k space), allowing the identification of multiple bonds having similar bond distances. ,

4.

(a) Wavelet-transformed XAS (WT-XAS) of Cu, Co, Ni, and Fe K-edge spectra of r-MnFeCoNiCuO _x and (b) Cu, Co, Ni, and Zn K-edge spectra of r-MnCoNiCuZnO _x . The references including metallic foils of Cu, Co, Ni, Fe, and ZnO were included.

For r-MnFeCoNiCuO _x , the peaks of Fe–M (R ≈2.1 Å and k ≈7.5 cm^–1), Co–M (R ≈2.1 Å and k ≈7.0 cm^–1), Ni–M (R ≈2.1 Å and k ≈7.2 cm^–1), and Cu–M (R ≈2.2 Å and k ≈7.3 cm^–1) (M = metallic Fe, Co, Ni, and Cu) were close to their respective references, i.e., Fe–Fe (R ≈2.2 Å and k ≈7.6 cm^–1), Co–Co (R ≈2.1 Å and k ≈6.7 cm^–1), Ni–Ni (R ≈2.2 Å and k ≈7.4 cm^–1), and Cu–Cu (R ≈2.2 Å and k ≈6.8 cm^–1). These slight deviations in R and k values underlined that the exsolved Fe, Co, Ni, and Cu exist in alloyed rather than pure elemental states. The same could be stated for r-MnCoNiCuZnO _x besides Zn, which possessed an oxide form like the pattern of ZnO.

3.3. Activity Test

Figure a exhibits the RWGS activity test at 350–500 °C. CO₂ conversions increased from 20.1% ± 3.0% to 43.5% ± 2.8% for MnFeCoNiCuO _x and 13.9% ± 2.8% to 38.3% ± 2.4% for MnCoNiCuZnO _x . CO dominated (>96%) in the tested temperatures. In 400 °C, CH₄ (3.9% and 2.0%) was detected for MnCoNiCuZnO _x and MnFeCoNiCuO _x while no CH₄ was detected above 450 °C for both catalysts. A close inspection can find a jump of CO₂ conversion in 400 (27.4%)–450 °C (39.6%) for MnFeCoNiCuO _x . Pure ZnO was tested under RWGS conditions and found inactive. Figure b shows the 100 h durability test for both catalysts at 450 °C. The conversions (MnFeCoNiCuO _x = 40.0% and MnCoNiCuZnO _x = 29.2%) were nearly unchanged with only CO as the detectable product.

5.

(a) Activity evaluation in 350–500 °C and (b) 100 h durability test for MnFeCoNiCuO _x and MnCoNiCuZnO _x at 450 °C. Reaction conditions: P = 0.1 MPa, H₂/CO₂ = 3, and GHSV = 40,000 mL/g_cat/h.

Figure S3 presents the in situ DRIFTS results under CO₂–H₂ switching. In the CO₂ stream, both catalysts exhibited characteristic bands at 1590 and 1300 cm^–1 (Δν ≈290 cm^–1), along with a minor band at 1040 cm^–1, attributable to bidentate chelating carbonate (b-*CO₃). , These b-*CO₃ bands intensified with increasing temperature from 50 to 350 °C. Upon switching to H₂ at 350 °C, the b-*CO₃ bands progressively diminished over time, accompanied by the appearance of gaseous CO bands at 2176 and 2113 cm^–1. Simultaneously, bands at 1540 and 1340 cm^–1 with a reduced Δν of ∼200 cm^–1 indicated the presence of residual b-*CO₃ due to varying basicity on the catalyst surface, and the new band at 1475 cm^–1 indicated the formation of surface bicarbonate (*HCO₃), likely due to partial reduction of b-*CO₃ in H₂.

3.4. Redox Property Evaluation

To investigate exsolution behavior, in situ XAS measurements were performed on the Fe K-edge of MnFeCoNiCuO _x and the Zn K-edge of MnCoNiCuZnO _x during H₂–CO₂ switching tests (see Figure ).

6.

In situ XAS spectra of Fe K-edge of MnFeCoNiCuO _x in (a) H₂ and (b) CO₂ at room temperature and 350, 450, and 500 °C and (b) Zn K-edge of MnCoNiCuZnO _x at room temperature and 450 °C in (c) H₂ and (d) CO₂.

For MnFeCoNiCuO _x under H₂ at room temperature, the Fe K-edge spectrum exhibited a strong white line with an edge at 7126.1 eV. As the temperature increased to 350–450 °C, both the white line intensity and edge position decreased, shifting to 7112.1 eV. Linear combination analysis indicated a reduction in the Fe oxidation state from +2.7 to +0.2 (see Figure S4a). Upon exposure to CO₂, the Fe edge showed the reverse trend: an increase in white line intensity and an upward shift in edge energy (from 7112.1 eV at room temperature to 7126.1 eV at 450 °C), with the Fe oxidation state increasing from +0.2 to +2.7 (see Figure S4b). These results demonstrate a reversible redox process of Feexsolution in H₂ and redissolution in CO₂characteristic of MnFeCoNiCuO _x .

In contrast, at 450 °C under H₂, the Zn K-edge spectrum of r-MnCoNiCuZnO _x evolves to closely match that of ZnO, confirming the absence of Zn–metal bonding and demonstrating that Zn²⁺ remains excluded from the exsolved alloy NPs. However, upon switching to CO₂ at 450 °C, the spectrum remained unchanged, indicating that Zn²⁺ does not reincorporate into the spinel units of MnCoNiCuZnO _x . This confirms the redox-inactive nature of Zn in MnCoNiCuZnO _x under the tested conditions.

Figure S2 further confirms the exsolution in H₂ and redissolution in CO₂ through XRD analysis. The CO₂ reoxidized r-MnFeCoNiCuO _x and r-MnCoNiCuZnO _x exhibited similar spinel diffraction patterns to their pristine forms, albeit with reduced diffraction peak intensities.

4. Discussion

4.1. Redox-Driven Structural Dynamics

This study highlights the critical role of cation identity in dictating redox-responsive exsolution within spinel-type HEOs under RWGS conditions. By examining MnFeCoNiCuO _x and MnCoNiCuZnO _x , both successfully synthesized as single-phase, entropy-stabilized spinels containing five cationswe maintain nearly identical configurational entropy (ΔS _config. = 12.4 and 12.3 J/mol/K, respectively), enabling an assessment of how the substitution of redox-active Fe³⁺ versus redox-inert Zn²⁺ alters redox dynamics and catalytic behaviors.

In situ XAS and WT-XAS analyses confirm that MnFeCoNiCuO _x undergoes reversible exsolution and redissolution of Fe/Co/N/Cu NPs alloy under H₂–CO₂ cycles. In contrast, MnCoNiCuZnO _x shows limited redox responsiveness, with Zn²⁺ resisting the redox transformation. A possible explanation of Zn’s redox inertness may be due to its high standard Gibbs free energy of reduction (ΔG° > 50 kJ/mol). Figure illustrates the redox dynamics of MnFeCoNiCuO _x and MnCoNiCuZnO _x .

7.

The scheme of reversible exsolution and redissolution of Fe/Co/Ni/Cu NPs alloy on MnFeCoNiCuO _x and Co/Ni/Cu NPs alloy on MnCoNiCuZnO _x with the change of the local coordination of Zn²⁺.

4.2. Surface Oxygen Vacancies and RWGS Activity

The enhanced catalytic performance of MnFeCoNiCuO _x correlates with its higher surface O_ν concentration, as indicated by an increased O^β/O^α ratio and greater H₂ uptake relative to MnCoNiCuZnO _x . Oxygen vacancies play a role in facilitating CO₂ activation, consistent with previous findings in other HEOs such as Zr_0.5(NiFeCuMnCo)_0.5O _x and CoNiFeZnCrO _x . In our system, the onset of high RWGS activity above 400 °C in MnFeCoNiCuO _x is accompanied by the formation of Fe/Co/Ni/Cu alloy NPs, mirroring trends observed in La­(FeCoNiCrMn)­O₃ perovskites and (CrMnFeCoNiCuZn)₃O₄ spinels, where exsolved NPs and elevated O_ν concentrations lower light-off temperatures in CO oxidation. Furthermore, the high CO selectivity and durability of MnFeCoNiCuO _x under moderate conditions underscores its industrial promise of RWGS, especially in processes coupling renewable hydrogen with CO₂ to produce syngas for downstream value-added products.

Complementary CO₂-TPD and in situ DRIFTS analyses further support these observations. MnFeCoNiCuO _x exhibits greater CO₂ adsorption capacity in the reduced state, with bidentate carbonate (b-*CO₃) identified as the dominant intermediate formed via chemisorption at O_ν sites. , Under H₂ flow, the transient intensification and subsequent decline of b-*CO₃ signals suggest a RWGS mechanism involving carbonate formation followed by hydrogenationlikely catalyzed by the exsolved Fe/Co/Ni/Cu NPs. ,

4.3. Implications for Entropy-Stabilized Catalyst Design

While configurational entropy supports phase stability in HEOs, it does not guarantee redox adaptability or catalytic regeneration. The presence of redox-flexible cations, such as Fe³⁺, is essential for enabling dynamic structural evolution and reversible alloy formation. In contrast, redox-inert Zn²⁺ dilutes active sites and is inactive in CO₂ activation, consistent with previous reports of its inactivity in standalone catalytic conversion. , Thus, effective catalyst design should integrate entropy stabilization with the selection of redox-active components to tailor structural dynamics and reaction performance.

Spinel-type HEOs have proven their value in real-world catalysis, offering exceptional thermal stability, resistance to deactivation (e.g., anticoking and antisintering), and dynamic active-site behavior under harsh conditions. For example, a (CoCrFeNiAl)₃O₄ spinel HEO achieved over 80% H₂ yield with sustained performance in ethanol steam reforming at 600 °C, likely due to the self-reorganization and reversible metal exsolution. This underlines the structural resilience and redox agility required in industrial catalysts, reinforcing the practical relevance of our entropy-stabilized, redox-tunable HEOs.

5. Conclusions

This work reveals the important role of redox-active cation selection in enabling reversible exsolution–dissolution behavior in high-entropy spinel oxides for RWGS catalysis. MnFeCoNiCuO _x demonstrates dynamic redox adaptability through the reversible formation and redissolution of Fe/Co/Ni/Cu alloy NPs under H₂–CO₂ cyclingbehavior absent in the Zn-containing MnCoNiCuZnO _x . The enhanced RWGS activity of MnFeCoNiCuO _x , particularly above 400 °C, is attributed to the synergy of oxygen vacancies and exsolved Fe/Co/Ni/Cu alloy NPs. Notably, a similar configurational entropy in both systems does not guarantee redox responsiveness; rather, the presence of reducible cations is essential for catalytic regeneration. These findings offer a rational strategy for designing advanced HEOs catalysts by coupling entropy stabilization with targeted redox-active cation engineering.

Data will be made available on request.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.5c14624.

• N₂ isotherms and porosities; XPS-estimated relative surface cation composition; XRD patterns of MnFeCoNiCuO _x , r-MnFeCoNiCuO _x , and reoxidized r-MnFeCoNiCuO _x ; MnCoNiCuZnO _x , r-MnCoNiCuZnO _x , and reoxidized r-MnCoNiCuZnO _x ; in situ DRIFT analysis of a CO₂–H₂ switching test; and first derivatives of Fe XAS K-edge spectra of MnFeCoNiCuO _x in H₂ and CO₂ streams (PDF)

Pei-Tung Chou: Investigation, formal analysis, and data curation. Cheng-Chia Kuo: Investigation, formal analysis, and data curation. Po-Yang Peng: Investigation, formal analysis, and data curation. Ying-Rui Lu: Investigation, formal analysis, and data curation. Chi-Liang Chen: Investigation, formal analysis, data curation, and resources. Yu-Chuan Lin: Writingoriginal draft and review and editing, supervision, project administration, methodology, funding acquisition, and conceptualization.

The authors declare no competing financial interest.