Selective CO₂ hydrogenation enhanced by tuning the zinc content in nickel catalysts

Abstract

Electron-rich Ni sites in Ni₃Zn–Al₂O₃ drive CO production through monodentate formate decomposition. Meanwhile, a Zn-evaporation-mediated strategy was proposed to tune Zn content, and engineered electron-deficient Ni–Al₂O₃ promotes CH₄ formation by enabling bidentate formate hydrogenation with abundant *H under lean redox conditions (CO₂ : H₂ = 1 : 1).

---

New concepts

Achieving tunable selectivity in CO₂ hydrogenation remains challenging and requires precise control over the electronic structure of Ni. Herein, we developed a zinc-evaporation-mediated strategy, which tuned the Zn content and transformed a pristine Ni₃Zn–Al₂O₃ alloy catalyst into a Ni–Al₂O₃ dealloyed catalyst. This structural evolution switches the hydrogenation selectivity from CO on the alloy to CH₄ on the dealloyed catalyst. The distinct selectivity originates from the electronic differences at the Ni sites. Ni within the Ni₃Zn–Al₂O₃ alloy exhibited an electron-rich character at the metal–oxide interface, whereas Ni in the dealloyed Ni–Al₂O₃ featured an electron-deficient structure at the oxide–metal interface. These distinct electronic states critically dictated the activation and adsorption of the key reaction intermediates in the reaction. While both catalysts followed an associative pathway for CO₂ hydrogenation, monodentate formate decomposition dominates in the alloyed catalyst, whereas the dealloyed system favors bidentate formate hydrogenation to CH₄.

1. Introduction

The catalytic hydrogenation of CO₂ using renewable H₂ offers a promising route to mitigate anthropogenic climate change while valorizing carbon resources.^1,2 Nickel-based catalysts are particularly attractive for this process due to their high activity, low cost and abundance relative to noble metals.^1,3 However, achieving tunable selectivity between the competitive pathways towards CO via the reverse water–gas shift (RWGS) reaction and CH₄ via methanation remains a significant challenge under ambient pressure conditions.

Product selectivity hinges critically on the catalyst's ability to control C–O bond scission, intermediate adsorption strength, and hydrogenation ability.^4–6 This is governed by the electronic structure of the active Ni sites and their local microenvironment.⁵ Electron-rich Ni sites typically favor CO formation by weakening CO adsorption and facilitating desorption.⁷ Conversely, electron-deficient Ni sites promote H₂ activation, deeper hydrogenation of intermediates, and consequently, CH₄ production.^8–10 For instance, while alloying Ni with electronegative metals (e.g., Ru) can create electron-deficient sites favoring CH₄,¹⁰ pairing Ni with less electronegative metals like Zn (Pauling scale: Zn = 1.65, Ni = 1.91) can generate electron-rich Ni sites (Ni^δ−–Zn^δ+), suppressing CO adsorption and H₂ dissociation to favor CO.^11,12

Notably, Zn exhibits high volatility (m.p. 419.5 °C) and weak metal bonding, enabling its migration from alloys at elevated temperatures.^13–15 Recent studies exploit this phenomenon for catalyst synthesis (e.g., intermetallic, porous materials)^16–19 and optimization (e.g., creating ZnO_x–metal interfaces, dual sites, dispersed Zn²⁺).^20–22 However, employing volatile component migration to deliberately modulate the electronic states of Ni and thereby switching CO₂ hydrogenation selectivity between CO and CH₄ remains unexplored.

Herein, we report a zinc-evaporation-mediated strategy that tunes the Zn content and dynamically transforms a pristine Ni₃Zn–Al₂O₃ alloy catalyst into a dealloyed Ni–Al₂O₃ material. This structural evolution, driven by controlled Zn migration, induces a profound switch in selectivity: from CO at the electron-rich Ni sites within the Ni₃Zn–Al₂O₃ alloy to CH₄ at the electron-deficient Ni sites within the dealloyed Ni–Al₂O₃ catalyst. We elucidate the origin of this tunable selectivity through the distinct electronic states at the Ni sites and their impact on intermediate activation and reaction pathways. Furthermore, we demonstrate outstanding activity and stability for CH₄ production under a challenging lean redox environment (CO₂ : H₂ = 1 : 1), a critical condition for practical application. This study establishes volatile element migration as a powerful tool for in situ electronic modulation and selectivity control in CO₂ hydrogenation.

2. Results and discussion

2.1. Catalytic performance of selectivity switching under lean redox conditions

In this work, we propose a zinc-evaporation-mediated strategy to design two nickel-based catalysts (Ni₃Zn–Al₂O₃ and Ni–Al₂O₃) for selective CO₂ hydrogenation. Initially, the CO₂ hydrogenation behavior of two catalysts was determined at 200–410 °C at a CO₂ : H₂ ratio of 1 : 1, and the results are shown in Fig. 1a and b. Both Ni₃Zn–Al₂O₃ and Ni–Al₂O₃ presented similar CO₂ conversion, which increased with the rise of the reaction temperature. As expected, the two catalysts had distinct selectivity, i.e., CO₂ was converted to major CO on Ni₃Zn–Al₂O₃, whereas CO₂ was hydrogenated to predominantly CH₄ on Ni–Al₂O₃. This is verified by the different activation energies (E_a) for CO and CH₄ formation on the two catalysts, as shown in Fig. 1e and f, where the E_a values of CO and CH₄ formation were 73.0 and 81.1 kJ mol⁻¹ on Ni₃Zn–Al₂O₃, and 95.6 and 39.9 kJ mol⁻¹ on Ni–Al₂O₃, respectively. Furthermore, a long-term catalytic test was carried out at 350 °C under identical reaction conditions. Both catalysts showed stable performance for 100 h. Specifically, Ni₃Zn–Al₂O₃ exhibited 14.0% CO₂ conversion with 91.6% CO selectivity, while Ni–Al₂O₃ achieved a stable CO₂ conversion of 19.5% with 77.7% CH₄ selectivity (Fig. 1c and d).

Fig. 1 (a) and (b) CO₂ hydrogenation activity tests under a CO₂ : H₂ ratio of 1 : 1, and (c) and (d) the stability tests of the CO₂ hydrogenation reaction at 350 °C under a CO₂ : H₂ ratio of 1 : 1 on (a) and (c) Ni₃Zn–Al₂O₃ and (b) and (d) Ni–Al₂O₃. (e) and (f) the Arrhenius plots for (e) CH₄ and (f) CO formation activation energy.

2.2. Zn evaporation-mediated structural transformation

For catalyst synthesis, the Ni₃Zn–Al₂O₃ catalyst was first obtained using a coprecipitation–hydrothermal method, as reported in detail in our previous work.²³ We have noticed that Zn exhibits high volatility (m.p. 419.5 °C) and weak metal bonding, enabling its migration from alloys at elevated temperatures.^13–15 Moreover, the migration of volatile components could modulate the electronic state of Ni and thereby switch CO₂ hydrogenation selectivity between CO and CH₄. Therefore, the Zn evaporation-mediated strategy was proposed to design Ni–Al₂O₃ catalysts for tailoring the selectivity of CO₂ hydrogenation from CO to CH₄ as shown in Fig. 2a. During the process, metallic Zn on Ni₃Zn–Al₂O₃ was evaporated from the alloy component, as evidenced by the presence of argent deposition on the quartz tube wall (Fig. S1). The XRD pattern of Ni₃Zn–Al₂O₃, shown in Fig. 2b, exhibits characteristic peaks at 43.7°, 50.9° and 75.3°, which are attributed to the (111), (200) and (220) planes of Ni₃Zn. The zinc evaporation-mediated Ni–Al₂O₃ showed a blue shift of sharp diffraction peaks at 44.5°, 51.8° and 76.4°, corresponding to the (111), (200) and (220) planes of face-centered cubic Ni in the XRD pattern (Fig. 2b). Moreover, the evaporation of volatile Zn metal in pristine catalysts from the original alloy architectures results in increased loading of the remaining active metal. The element analysis by ICP–OES, shown in Table S1, confirmed the increase in Ni loading from 52.0 wt% to 87.7 wt% and decrease in Zn loading from 30.0 wt% to 3.3 wt% before and after Zn migration.

Fig. 2 (a) Schematic diagram of the Zn evaporation-mediated strategy for the conversion of the Ni₃Zn alloy catalyst to Ni–Al₂O₃ under H₂ at 800 °C. (b) XRD patterns of fresh Ni₃Zn–Al₂O₃ and Ni–Al₂O₃. (c) In situ PXRD patterns for the structure evolution on pristine Ni₃Zn–Al₂O₃ under H₂ for 3 h.

To further enhance the understanding of structural evolution of the catalyst using the evaporation of Zn species, we performed in situ PXRD experiments under pure H₂ on pristine Ni₃Zn–Al₂O₃ at 800 °C, as depicted in Fig. 2c. The alloy sample was first obtained in the reaction chamber at 550 °C for 2 h under H₂, and then the temperature was increased to 800 °C. The three diffraction peaks for the Ni₃Zn alloy shifted toward higher 2θ values, accompanied by an increase in temperature to 800 °C. Specifically, this peak shift was significant in the first 1 h, indicating the fast evaporation of Zn from the alloy during the initial period. This trend slowly progressed with further increases in treatment time. This observation suggested the occurrence of the migration process through Ni₃Zn(s) → Ni₃(s) + Zn(g) during reduction under high-temperature conditions. Concurrently, the diffraction peaks of Ni–Al₂O₃ became narrower and sharper in comparison with the peaks on Ni₃Zn–Al₂O₃, suggesting the growth in Ni particle size on the former.

2.3. Morphological and interface evolution

To attain a more comprehensive understanding of the structural nature of Ni₃Zn–Al₂O₃ and Ni–Al₂O₃, HAADF–STEM and the corresponding EDS elemental analysis were carried out, and the results are presented in Fig. 3. The average particle size of fresh Ni₃Zn–Al₂O₃ and Ni–Al₂O₃ was 10.8 and 40.4 nm, respectively, derived from TEM images shown in Fig. S2. The larger particle size of Ni–Al₂O₃ was found to originate from particle aggregation during high temperature treatment. Notably, it is well-established that Ni particles larger than 8 nm generally favor CH₄ formation, while those below 5 nm promote CO production.^24–26 Since both catalysts studied here had Ni particles with sizes larger than 10 nm, yet exhibited markedly different product selectivities, particle size is unlikely to be the determining factor. For Ni₃Zn–Al₂O₃, the HAADF–STEM image exhibits lattice spacings of 0.206 and 0.180 nm in Fig. 3a, corresponding to the (111) and (220) fringes of Ni₃Zn, respectively. The elements of Al (orange color) and O (red color) of Al₂O₃ support were well overlapped, while the elements of Ni (blue color) and Zn (yellow color) were homogeneously mixed and dispersed on Al₂O₃ (Fig. 3b and c). In contrast, for Ni–Al₂O₃, shown in Fig. 3d–f, the HAADF–STEM images displayed lattice fringes of 0.204 and 0.179 nm for the (111) and (200) planes of Ni. EDS mapping analysis confirms the segregation of Zn from Ni in the bulk structure, and the weak Zn signal suggests complete Zn migration in 10 h. Moreover, Al and O partially covered and deposited on the exterior surface of large Ni particles in the form of Al₂O₃ islands. The TEM images of other regions of the Ni–Al₂O₃ catalyst in Fig. S3 give the intuitive impression that Al₂O₃ islands covered the metallic Ni exterior surface, which is different from Ni₃Zn–Al₂O₃, where small alloy particles were deposited on the Al₂O₃ support surface. The resultant Ni–Al₂O₃ catalyst exhibits an inverse structure, wherein the active metal acts as a support for oxide domains. This architecture differs fundamentally from conventional supported catalysts, where metal nanoparticles are dispersed on an oxide support. In inverse catalysts, oxide islands are dispersed across a continuous metal matrix, which maximizes the perimeter and density of the metal–oxide interfacial sites.²⁷ These interfacial sites facilitate the adsorption of active hydrogen species and help optimize the surface coverage of reactants and intermediates.^28,29 The synergistic interactions between the metal and oxide domains enhance the hydrogenation of surface species, as a feature particularly favorable for CH₄ formation observed in our experiments.

Fig. 3 (a), (b), (d) and (e) HAADF–STEM images and (c) and (f) the corresponding EDS elemental mappings of (a)–(c) Ni₃Zn–Al₂O₃ and (d)–(f) Ni–Al₂O₃.

The nature of active sites prominently affects the hydrogenation selectivity. To investigate changes in the active sites derived by structure reconstruction, temperature programmed surface reactions of CO₂ hydrogenation under a CO₂ : H₂ ratio of 1 : 1 on Ni₃Zn–Al₂O₃ and Ni–Al₂O₃ were carried out to disclose the different activation behaviors for the reactant (Fig. 4a and b). The on-set temperature for CO₂ conversion on the two catalysts is almost the same, while completely different products are obtained, i.e., the first formation of CO at 184 °C on Ni₃Zn–Al₂O₃, but of CH₄ at 170 °C on Ni–Al₂O₃. In addition, the formation of CH₄ on Ni₃Zn–Al₂O₃ and CO on Ni–Al₂O₃ starts at 280 and 205 °C, respectively. Notably, CO formation was predominated by CH₄ formation on Ni₃Zn–Al₂O₃ at a temperature below 400 °C, whereas CH₄ is generated as the major product instead of CO on Ni–Al₂O₃ at low temperatures (<500 °C). The CO pulse chemisorption measurement revealed that the active surface areas of Ni₃Zn–Al₂O₃ and Ni–Al₂O₃ were 38.0 and 17.4 m² g⁻¹, respectively. The relatively lower active surface area of Ni–Al₂O₃ is probably due to the presence of oxide islands partially covering the agglomerated Ni surface.²⁷ The similar onset temperatures of CO₂ conversion indicated that the exposed Ni sites on Ni–Al₂O₃ exhibit higher intrinsic activity. This demonstrates that the nature of active sites on Ni₃Zn–Al₂O₃ and Ni–Al₂O₃ changed, which is related to the difference in product selectivity.

Fig. 4 (a) and (b) TPSR curves of CO₂ and H₂ with a ratio of 1 : 1 on (a) Ni₃Zn–Al₂O₃ and (b) Ni–Al₂O₃.

2.4. Electronic structure differences of Ni₃Zn–Al₂O₃ and Ni–Al₂O₃

Alloying has generally been proposed to modulate electronic structure. Bader charge and the deformation charge density calculations confirm the electron transfer from Zn to Ni in the ZnNi₃ alloy, leading to the formation of electron-rich Ni sites.¹² To probe surface electronic states, XPS measurements were conducted (Fig. 5a and b and Table S2). The Ni 2p_3/2 spectrum of Ni₃Zn–Al₂O₃ exhibited binding energies at 851.8 (Ni⁰), 854.8 (Ni²⁺), and 861.0 (satellite) eV, whereas those for Ni–Al₂O₃ showed distinct positive shifts to 852.5 (Ni⁰), 855.7 (Ni²⁺), and 861.3 eV (satellite). The presence of NiO on the surface originated from the air exposure during handling. Critically, the 0.7 eV higher binding energy for Ni⁰ in Ni–Al₂O₃ signifies electron-deficient Ni sites.

Fig. 5 XPS spectra of (a) Ni 2p_3/2, and (b) O 1s of Ni₃Zn–Al₂O₃ and Ni–Al₂O₃. (c) H₂-TPD profiles of Ni₃Zn–Al₂O₃ and Ni–Al₂O₃.

This electronic contrast is further corroborated by the O 1s spectra (Fig. 5b). The binding energies of the O 1s spectrum of Ni₃Zn–Al₂O₃ at 530.4, 531.9 and 533.3 eV are attributed to the lattice oxygen (O_lat.), oxygen vacancies (O_vac.) and surface adsorbed oxygen species (O_ads.), while the corresponding peaks on Ni–Al₂O₃ appeared at 530.2, 531.7 and 533.1 eV.^30–34 Compared to Ni₃Zn–Al₂O₃, the adsorbed oxygen species remained unchanged on Ni–Al₂O₃, but the oxygen vacancies increased from 50% to 60%, accompanied by a decrease in lattice oxygen species from 18% to 5%. Oxygen vacancies serve as electron traps,^9,35 facilitating electron transfer from Ni to oxygen. Compared to Ni₃Zn–Al₂O₃, the concurrent blue shift in Ni 2p_3/2 and red shift in O 1s binding energies on Ni–Al₂O₃ unambiguously demonstrated charge transfer from Ni to oxygen, forming electron-deficient Ni sites and confirming strong electronic interactions between Ni and Al₂O₃.

To elaborate on the different H₂ activation abilities of Ni sites on the two catalysts, H₂-TPD analysis was carried out as shown in Fig. 5c. Both catalysts exhibited two peaks at 165 and 277 °C for Ni₃Zn–Al₂O₃ and at 242 and 440 °C for Ni–Al₂O₃, respectively. The peaks at deposition temperatures below 200 °C and in the range of 200–300 °C were attributed to the presence of weakly and strongly adsorbed H₂ on the top sites of metallic Ni⁰ species,^36,37 while the peak at a temperature above 400 °C is ascribed to the interface adsorption of hydrogen dissociated from electron-deficient Ni sites.^8,9,38 Enhanced H₂ activation on Ni–Al₂O₃ created a hydrogen-rich environment, promoting significant hydrogenation abilities.

The effect of the electronic structure of Ni on the adsorption/desorption of CO as the main intermediate was investigated at 25 °C using in situ DRIFTS. As shown in Fig. 6, Ni₃Zn–Al₂O₃ (Fig. 6a) and Ni–Al₂O₃ (Fig. 6b) presented CO adsorption peaks at 2170 and 2116 cm⁻¹ and in the range of 2000–2050 cm⁻¹, corresponding to gaseous CO and linearly adsorbed CO, respectively. Unlike Ni₃Zn–Al₂O₃, Ni–Al₂O₃ exhibited an apparent shoulder peak at 1970 cm⁻¹, which is ascribed to the bridged CO adsorbed on Ni flat planes in an oxide–metal interface.^39,40 Notably, the intensity of linearly and bridged CO on Ni–Al₂O₃ was significantly higher than that on Ni₃Zn–Al₂O₃, indicating the superior CO adsorption strength on the former. This is further confirmed by the relatively high adsorption strength of CO species on Ni–Al₂O₃ in comparison with that on Ni₃Zn–Al₂O₃ under Ar purging for 30 min. This matches well with density functional theory (DFT) studies of CO adsorption energy on Ni(111) and NiZn(111) surfaces.⁴¹ On the NiZn(111) surface, CO was adsorbed on the electron-rich h-Ni₂ site with an adsorption energy of –1.68 eV. However, on the Ni(111) surface, CO was adsorbed stably on the h-Ni₃ site via bonding with the C atom, with an enhanced adsorption energy of –2.16 eV.

Fig. 6 In situ DRIFTS of CO adsorption and desorption on (a) Ni₃Zn–Al₂O₃ and (b) Ni–Al₂O₃ at 25 °C. The reaction times for the spectra shown are 1, 2, 3, 5, 10, 15, 20, 25, and 30 min from bottom to top for adsorption and desorption, respectively.

Generally, CO chemisorption on metals involves two primary mechanisms: σ donation, where electrons transfer from the CO(5σ) orbital into empty metal bands, and π back-donation, where electrons transfer from occupied metal d-bands into the CO(2π) orbitals. Thermochemically, π back-donation dominates the bonding interaction.^42,43 Ab initio self-consistent field calculations for the Ni/Zn system reveal that Zn and Ni interactions caused Ni d electrons to shift towards the bimetallic bond region (metal–metal interface).⁴⁴ This elevates the Ni d orbital energy (away from the CO(2π) level), rendering these electrons unavailable for effective π back-donation into CO. Consequently, Zn to Ni electron transfer weakens π back-donation, leading to attenuated CO adsorption strength on the Ni₃Zn alloy.

Conversely, Zn migration eliminates this charge transfer effect in the dealloyed Ni–Al₂O₃ catalyst. The absence of electron density redistribution within the former bimetallic bonding region enhances d electron back-donation towards CO, strengthening the Ni–C bond and resulting in significantly stronger CO adsorption. This fundamental difference in adsorption behavior directly dictates product selectivity: facile CO desorption from the electron-rich Ni sites in Ni₃Zn–Al₂O₃ favors high CO production, while persistent CO chemisorption on the electron-deficient Ni sites in Ni–Al₂O₃ suppresses CO release and enables further hydrogenation toward CH₄.

2.5. The reaction pathway for CO₂ hydrogenation on Ni catalysts with different electronic structure

To illustrate the effect of electronic structure in Ni catalysts on reaction pathways, in situ DRIFTS measurements were then carried out to study adsorption behavior, identify the surface intermediates, and monitor the reaction mechanism of CO₂ adsorption (Fig. S4) and subsequent hydrogenation with a CO₂ : H₂ ratio of 1 : 1 over Ni₃Zn–Al₂O₃ and Ni–Al₂O₃ (Fig. 7). During CO₂ adsorption (Fig. S4), both catalysts exhibited peaks at 2200–2400 and 1510/1360 cm⁻¹, which were assigned to the presence of gaseous CO₂ and monodentate carbonate species, respectively.^45–47 Besides, Ni–Al₂O₃ presented an extra peak at 1645 cm⁻¹, corresponding to the formation of bicarbonate species.^48,49 No peaks for CO intermediates, such as linearly or bridged CO, were observed on either catalyst. This suggests that the direct CO₂ dissociation pathway cannot occur during CO₂ adsorption on the two catalysts.

Fig. 7 The in situ DRIFTS spectra of (a) and (b) Ni₃Zn–Al₂O₃ and (d) and (e) Ni–Al₂O₃ for CO₂ hydrogenation. Symbols: (▼) CH₄; (▲) gaseous CO; (○) bicarbonate; (Δ) monodentate formate; (●) bidentate formate; and (□) monodentate carbonate. The reaction times for the spectra shown are 1, 2, 3, 5, 10, 15, 20, 25, and 30 min from bottom to top. (c) and (f) Schematic illustrations of different reaction pathways on (c) Ni₃Zn–Al₂O₃ and (f) Ni–Al₂O₃ for CO₂ hydrogenation.

Upon introducing H₂ into the reaction cell, in situ DRIFTS studies for CO₂ hydrogenation were performed. For Ni₃Zn–Al₂O₃, the shoulder peak for monodentate formate (m-*HCOO⁻) appeared at 1590/1280 cm⁻¹ after 5 min, originating from the hydrogenation of monodentate carbonate species.^50,51 Concurrently, gaseous CO (2170 and 2116 cm⁻¹) and CH₄ (3016 cm⁻¹) are formed on the control catalyst, as shown in Fig. 7a and b.⁴⁵ The predominate formation of CO over CH₄ was attributed to the poor hydrogenation ability of formate intermediates on Ni₃Zn–Al₂O₃. In contrast, Ni–Al₂O₃ not only produced gaseous CH₄ and CO, but also had new peaks presented at 2952 cm⁻¹ and 1580/1540 cm⁻¹, corresponding to the formation of bidentate formate intermediates (b-*HCOO⁻, Fig. 7d and e), which can be derived from the hydrogenation of bicarbonate species.^34,52

In situ DRIFTS reveals two distinct associative pathways for CO₂ hydrogenation (Fig. 7c and f). While both catalysts initially chemisorb CO₂ forming carbonate species, the divergent electronic structures of Ni sites dictated subsequent intermediate evolution. The Ni₃Zn–Al₂O₃ catalyst, featuring small particles with abundant unsaturated sites, stabilizes monodentate formate (m-*HCOO, 1590/1280 cm⁻¹). This configuration, as evidenced by coordination through a single O atom, exhibits lower thermodynamic stability and feasible formation (0.59 eV less stable than bidentate formate and 27 kJ mol⁻¹ lower than the standard enthalpy of the corresponding formation),^53,54 favoring rapid decomposition to CO (Fig. 7c). Conversely, Zn migration drives particle agglomeration in Ni–Al₂O₃, creating extended flat facets that stabilize bidentate formate (b-*HCOO, 1580/1540 cm⁻¹) via dual O-atom coordination to Ni atop sites.⁵² This pathway proceeds through bicarbonate intermediates (1645 cm⁻¹), formed via sequential hydrogenation of monodentate carbonate. According to XPS analysis, we concluded that Zn evaporation caused an increase in oxygen vacancy defects. In metal oxides, the electronegativity of oxygen is generally higher than that of other elements. Therefore, after oxygen atoms are removed, oxygen vacancies display positive charge states,^55,56 with the ability to capture electrons. The electron transfer from metal to oxide not only induced the formation of electron-deficient Ni, which facilitated hydrogen activation and dissociation, but it also promoted oxide–metal interaction and resulted in an oxide–metal interface with more effective adsorption sites for *H species. Moreover, the inverse structure (Al₂O₃ islands on Ni particles) provides abundant oxide–metal interfaces, enriching the surface with dissociated *H species. These readily participate in stepwise hydrogenation, converting stable bidentate formate to CH₄ (Fig. 7f), which is consistent with the reported kinetics under *H-rich conditions.^57,58 Therefore, Zn alloying in Ni₃Zn–Al₂O₃ promotes CO desorption and limits reactive *H coverage, favoring RWGS. The dealloyed Ni–Al₂O₃, however, leverages its restructured geometry to stabilize bidentate formate and maximize interfacial *H availability, collectively reducing the energetic barrier for deep hydrogenation to methane.

3. Conclusions

This work establishes volatile-element migration as a transformative strategy for in situ catalyst restructuring, demonstrating how controlled Zn evaporation dynamically transforms the Ni₃Zn–Al₂O₃ alloy into a dealloyed Ni–Al₂O₃ catalyst to achieve a complete selectivity switch from CO to CH₄ in CO₂ hydrogenation. The electron-rich Ni sites and fragmented surfaces in Ni₃Zn–Al₂O₃ favor unstable monodentate formate formation and facile CO desorption, while Zn removal generates an inverse structure (Al₂O₃ islands on enlarged Ni particles) with electron-deficient Ni sites that enhance H₂ dissociation, enrich interfacial *H coverage, and stabilize bidentate formate on sufficient interface oxygen vacancies for deep hydrogenation to CH₄ even under lean H₂ conditions. These findings underscore how targeted in situ restructuring mediates both electronic states (Ni^δ− → Ni^δ+) and geometric configurations to dictate reaction pathways, offering a generalizable route for dynamic catalyst optimization in CO₂ valorization.

Author contributions

M. Cao: writing – original draft, investigation, and formal analysis. Y. Huang: methodology, investigation. Y. Gao: formal analysis. Z. Wang: formal analysis. Q. Wang: methodology and formal analysis. S. Li: project administration. F. Yu: formal analysis. L. Qiu: formal analysis. R. Li: resources and funding acquisition. X. Yan: writing – review and editing, conceptualization, and funding acquisition. Y.-X. Pan: supervision, conceptualization and funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included within the article and supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5nh00736d.

The data supporting the findings of this study are available from the corresponding author (yxpan81@sjtu.edu.cn) upon reasonable request.

Acknowledgements

The authors acknowledge the National Natural Science Foundation of China (No. 22278286) and the Science Foundation for Distinguished Young Scholar of Shanxi Province (202303021223001). Dr Yan also thanks Dr Geoffrey A. Ozin from the University of Toronto for their useful discussion of the dealloying process.

References

1. R. Ye, J. Ding, T. Reina, M. Duyar, H. Li, W. Luo, R. Zhang, M. Fan, G. Feng, J. Sun and J. Liu, Nat. Synth., 2025, 4, 288–302.
2. Z. Liu, Z. Deng, S. Davis and P. Ciais, Nat. Rev. Earth Environ., 2024, 5, 253–254.
3. F. Hu, R. Ye, Z. Lu, R. Zhang and G. Feng, Energy Fuels, 2022, 36, 156–169.
4. W. Zhang, T. Pu, Z. Wang, L. Shen and M. Zhu, Ind. Eng. Chem. Res., 2022, 61, 9678–9685.
5. J. Hu, Y. Cai, J. Xie, D. Hou, L. Yu and D. Deng, Chemistry, 2024, 10, 1084–1117.
6. D. Gao, W. Li, H. Wang, G. Wang and R. Cai, Trans. Tianjin Univ., 2022, 28, 245–264.
7. S. Zhang, H. Ma, L. Jia, Z. Zhang, X. Li, S. Dang, Y. Huang, Y. Tian, W. Tu and Y. Han, Appl. Catal., B, 2025, 361, 124646.
8. H. Li, J. Zhang, Y. Liu, L. Zheng, X. Cao, Y. He and D. Li, Chem. Eng. Sci., 2022, 260, 117867.
9. Y. Zhao, T. Zhu, J. Li, C. Li, W. Lu, J. Liu, D. Zhang, Y. Dong and M. Yang, ACS Catal., 2025, 15, 5490–5502.
10. X. Yan, M. Cao, S. Li, P. Duchesne, W. Sun, C. Mao, R. Song, Z. Lu, X. Chen, W. Qian, R. Li, L. Wang and G. Ozin, J. Am. Chem. Soc., 2023, 145, 27358–27366.
11. L. Shen, W. Zhang, Y. Feng, J. Xu and M. Zhu, J. Mater. Chem. A, 2023, 11, 8248–8255.
12. S. Lin, Q. Wang, M. Li, Z. Hao, Y. Pan, X. Han, X. Chang, S. Huang, Z. Li and X. Ma, ACS Catal., 2022, 12, 3346–3356.
13. H. Sasaki, T. Nagai and M. Maeda, J. Alloys Compd., 2010, 504, 475–478.
14. B. Yao, C. Xu, Y. Tang, Y. Du, S. Tan, S. Dai, G. Luo and Q. He, Nano Today, 2025, 63, 102746.
15. X. Wang, Q. Li, S. Feng, Y. Liu, C. Xu, Q. Zhou, H. Liu and Q. Yan, Trans. Tianjin Univ., 2025, 31, 131–144.
16. P. Wei, S. Chen, R. Luo, G. Sun, K. Wu, D. Fu, Z. Zhao, C. Pei, N. Yan and J. Gong, Nat. Commun., 2024, 15, 8157.
17. Z. Lu, C. Li, J. Han, F. Zhang, P. Liu, H. Wang, Z. Wang, C. Cheng, L. Chen, A. Hirata, T. Fujita, J. Erlebacher and M. Chen, Nat. Commun., 2018, 9, 276.
18. Y. Li, Q. Zhang, X. Zhao, H. Wu, X. Wang, Y. Zeng, Q. Chen, M. Chen and P. Liu, Adv. Funct. Mater., 2023, 33, 2214124.
19. Y. Xia, Z. Lu, J. Han, F. Zhang, D. Wei, K. Watanabe and M. Chen, Acta Mater., 2022, 238, 118210.
20. T. Song, R. Li, J. Wang, C. Dong, X. Feng, Y. Ning, R. Mu and Q. Fu, Angew. Chem., Int. Ed., 2024, 63, e202316888.
21. H. Jia, X. Feng, X. Du, L. Lin, R. Mu and Q. Fu, Angew. Chem., Int. Ed., 2025, 64, e202503319.
22. S. Almutairi, B. Mezari, P. Magusin, E. Pidko and E. Hensen, ACS Catal., 2012, 2, 71–83.
23. Q. Wang, W. Wang, M. Cao, S. Li, P. Wang, J. He, R. Li and X. Yan, Appl. Catal., B, 2022, 317, 121806.
24. K. Feng, J. Tian, M. Guo, Y. Wang, S. Wang, Z. Wu, J. Zhang, L. He and B. Yan, Appl. Catal., B, 2021, 292, 120191.
25. J. Simons, T. de Heer, R. van de Poll, V. Muravev, N. Kosinov and E. Hensen, J. Am. Chem. Soc., 2023, 145, 20289–20301.
26. D. Adhikari, C. Whitcom, W. Zhang, S. Zhang and R. Davis, J. Catal., 2024, 438, 115708.
27. Y. Han, C. Yang, W. Shao, L. Cai, W. Wang, Z. Jin and C. Jia, Nat. Commun., 2019, 10, 3470.
28. C. Song, J. Liu, R. Wang, X. Tang, K. Wang, Z. Gao, M. Peng, H. Li, S. Yao, F. Yang, H. Lu, Z. Liao, X. Wen, D. Ma, X. Li and L. Lin, Nat. Chem. Eng., 2024, 1, 638–649.
29. H. Du, J. Fan, C. Miao, M. Gao, Y. Liu, D. Li and J. Feng, Trans. Tianjin Univ., 2021, 27, 24–41.
30. J. Tian, P. Zheng, T. Zhang, Z. Han, W. Xu, F. Gu, F. Wang, Z. Zhang, Z. Zhong, F. Su and G. Xu, Appl. Catal., B, 2023, 339, 123121.
31. Y. Wang, L. Li, G. Li, Q. Zhao, X. Wu, Y. Wang, Y. Sun and C. Hu, ACS Catal., 2023, 13, 6486–6496.
32. X. Jia, X. Zhang, N. Rui, X. Hu and C. Liu, Appl. Catal., B, 2019, 244, 159–169.
33. R. Ye, L. Ma, X. Hong, T. Reina, W. Luo, L. Kang, G. Feng, R. Zhang, M. Fan, R. Zhang and J. Liu, Angew. Chem., Int. Ed., 2024, 63, e202317669.
34. P. Hongmanorom, J. Ashok, P. Chirawatkul and S. Kawi, Appl. Catal., B, 2021, 297, 120454.
35. Y. Wang, X. Wang, M. Qiao, Q. Wu, K. Liu, X. Yi, R. Qin, G. Fu and N. Zheng, Sci. China: Chem., 2024, 67, 4134–4141.
36. R. Yan and W. Oscar, J. CO₂ Util., 2023, 68, 102381.
37. D. Wang, X. Kang, Y. Gu, H. Zhang, J. Liu, A. Wu, H. Yan, C. Tian and H. Fu, ACS Catal., 2020, 10, 10449–10458.
38. H. Zhao, C. Liu, Y. Zheng, S. Li, Y. Gao, Q. Ma, F. Wang and Z. Dong, ACS Catal., 2024, 14, 8619–8630.
39. D. Blackmond and E. Ko, J. Catal., 1985, 96, 210–221.
40. X. Zhu, P. Huo, Y. Zhang, D. Cheng and C. Liu, Appl. Catal., B, 2008, 81, 132–140.
41. P. Yin, H. Meng, L. Wang, Y. Lai, Y. Jie, J. Yu, W. Liu, X. Zhao, T. Shen, X. Zhang, J. Han, Y. Yang, H. Yan and M. Wei, J. Mater. Chem. A, 2022, 10, 16610–16619.
42. K. Hermann, P. Bagus and C. Nelin, J. Phys. Rev. B, 1987, 35, 9467.
43. E. Davidson, K. Kunze, F. Machado and S. Chakravorty, Acc. Chem. Res., 1993, 26, 628.
44. J. Rodriguez and M. Kuhn, J. Phys. Chem., 1996, 100, 381–389.
45. T. Zhang, W. Wang, F. Gu, W. Xu, J. Zhang, Z. Li, T. Zhu, G. Xu, Z. Zhong and F. Su, Appl. Catal., B, 2022, 312, 121385.
46. M. Akatsuka, A. Nakayama and M. Tamura, Appl. Catal., B, 2024, 343, 123538.
47. J. Di Cosimo, V. Díez, M. Xu, E. Iglesia and C. Apesteguía, J. Catal., 1998, 178, 499–510.
48. C. Morterra, A. Zecchina, S. Coluccia and A. Chiorino, J. Chem. Soc., Faraday Trans. 1, 1977, 73, 1544–1560.
49. H. Zhang, D. Mao, J. Zhang and D. Wu, Chem. Eng. J., 2023, 452, 139144.
50. S. Collins, M. Baltanás and A. Bonivardi, J. Catal., 2004, 226, 410–421.
51. A. Solis-Garcia, F. Louvier-Hernandez Jose, A. Almendarez-Camarillo and J. Fierro-Gonzalez, Appl. Catal., B, 2017, 218, 611–620.
52. Q. Pan, J. Peng, T. Sun, S. Wang and S. Wang, Catal. Commun., 2014, 45, 74–78.
53. G. Peng, S. Sibener, G. Schatz, S. Ceyer and M. Mavrikakis, J. Phys. Chem. C, 2012, 116, 3001–3006.
54. G. Ruehl, S. Harman, O. Gluth, D. LaVoy and C. Campbell, ACS Catal., 2022, 12, 10950–10960.
55. L. Dong, R. Jia, B. Xin, B. Peng and Y. Zhang, Sci. Rep., 2017, 7, 40160.
56. D. Liu, S. Clark and J. Robertson, Appl. Phys. Lett., 2010, 96, 032905.
57. C. Wang, Y. Lu, Y. Zhang, H. Fu, S. Sun, F. Li, Z. Duan, Z. Liu, C. Wu, Y. Wang, H. Sun and Z. Yan, Nano Res., 2023, 16, 12153–12164.
58. H. Kang, Y. Liu, Y. Lu, P. Zhang, M. Tang, Z. Gao, H. Ge and W. Fan, J. Catal., 2024, 432, 115427.

---