Nucleation and crystallization of metal oxides from carbonates

Abstract

Metal oxides play an important role in catalysis, integrated circuit fabrication, and optical coatings. A common approach for obtaining these oxides is the thermal decomposition of solid precursor templates, among which metal carbonates are the most widely used ones. Despite the importance of the process, the atomic-scale mechanisms governing the transformation of such templates into oxides remain insufficiently understood. Using in situ transmission electron microscopy (TEM), we show that La₂(CO₃)₃·8H₂O and Ce₂(CO₃)₃·8H₂O undergo a multistep transformation: (i) dehydrogenation, (ii) decarbonization, and (iii) crystallization. Our observations reveal concurrent compositional and structural changes within the precursor templates during their conversion to oxides. These findings offer valuable insights into the crystallization of metal oxides, providing a basis for optimizing their properties for diverse technological applications.

---

Introduction

Metal oxides such as CeO₂,^1,2 La₂O₃,^3,4 ZnO,^5,6 and Al₂O₃ ⁷ are widely used in catalysis,^1,5,7 electronics,³ and optics,⁴ due to their exceptional properties, including catalytic activity,^1,5 ionic conductivity,² high dielectric constant (k),³ and ultraviolet (UV) absorption.⁶ For instance, CeO₂ is a key component in automotive catalytic converters and solid oxide fuel cells,^1,2 while La₂O₃ is used as a high-k dielectric material in the manufacturing of integrated circuits.³ A major strategy for producing metal oxide nanostructures (e.g., particles,^8–10 plates,^11,12 and films¹³) is the thermal decomposition of solid or wet-coated precursor templates, such as metal (oxy)carbonates^8,9,12 and hydroxides.^10,11,14 This method enables precise control over the morphology,^8,9 composition,¹¹ and properties^9,11 of the resulting metal oxides.

The transformation process typically involves phase changes, often progressing through intermediate metastable phases, accompanied by the release of volatile byproducts, such as CO₂ and H₂O.^11,12,14 During this transformation, the crystallization of the oxide phase determines the crystal grain size,^12,15 composition,¹¹ and defect characteristics,¹⁶ which in turn influence the electronic,¹⁷ catalytic,^11,16 optical,¹⁸ and mechanical¹⁵ properties of the oxides.

Despite the widespread use of thermal decomposition for oxide synthesis, many details about the atomic-scale transformation remain unknown. This knowledge gap arises primarily from the lack of methods capable of directly probing the process. While previous studies based on macroscopic characterization methods, such as X-ray diffraction (XRD),^19–22 Fourier transform infrared spectroscopy (FTIR),^22,23 and thermogravimetric analysis (TGA),^22,23 have provided important insights into the transformations, they lack the resolution to capture atomic-scale nucleation pathways driving crystallization in real time. Transmission electron microscopy, although capable of atomic resolution, is conventionally performed under vacuum at room temperature, requiring sample cooling and transfer that interrupt ongoing reactions.^8,10,14,24 These constraints hinder the observation of rapid, short-lived events occurring on the timescale of seconds, leaving the early stages of structural evolution largely inaccessible.

Here, we employed in situ TEM imaging with atomic-scale resolution^20,25–29 to directly monitor the real-time transformation of metal carbonates into polycrystalline oxide nanostructures. This approach enables the simultaneous tracking of nucleation and phase evolution under controlled heating and gas environments, offering mechanistic insights that are beyond the reach of conventional methods. To investigate both intrinsic and environment-driven transformation processes, we used two types of metal carbonate templates: La₂(CO₃)₃·8H₂O and Ce₂(CO₃)₃·8H₂O. In the former, decomposition proceeds intrinsically to La₂O₃, with lanthanum retaining its +3 oxidation state, whereas in the latter, decomposition is oxygen-driven, leading to the formation of CeO₂, accompanied by the oxidation of cerium from Ce³⁺ to Ce⁴⁺.

Results and discussion

The as-synthesized La₂(CO₃)₃·8H₂O template is a plate-like crystal (Fig. 1A: t − t₀ = 0 s). It has a layered structure with an interlayer spacing of 8.6 Å (Fig. 1A and S1). Each basic layer consists of lanthanum polyhedra, in which La³⁺ ions are coordinated with CO₃²⁻ ions and water molecules. Approximately 75% of the total water molecules are coordinated within the polyhedra, while the remaining water molecules reside in the interlayer regions. Upon heating from 23 to 800 °C at a rate of 10 °C min⁻¹, the single-crystalline template decreased in size and transformed into polycrystalline La₂O₃ (Fig. 1A: t − t₀ = 4700 s). Similarly, when the Ce₂(CO₃)₃·8H₂O template was annealed under an oxygen atmosphere (760 Torr of 20% O₂ and 80% Ar) instead of a vacuum, it transformed into CeO₂ (Fig. 1B and S5), in a manner analogous to the transformation of La₂(CO₃)₃·8H₂O into La₂O₃.

Fig. 1 Transformations of carbonate templates into metal oxides. Schematics^12,30–32 and in situ TEM images with the corresponding electron diffraction patterns showing the transformation of single-crystalline (A) La₂(CO₃)₃·8H₂O and (B) Ce₂(CO₃)₃·8H₂O templates into polycrystalline La₂O₃ and CeO₂ plates, respectively, during heating. The La₂(CO₃)₃·8H₂O template was heated from 23 to 800 °C at a rate of 10 °C min⁻¹ in a vacuum, while the Ce₂(CO₃)₃·8H₂O template was heated from 23 to 600 °C at the same rate under an oxygen atmosphere (760 Torr of 20% O₂ and 80% Ar). For clarity, H atoms in H₂O molecules are omitted in the schematics. Black dashed rectangular prisms indicate the corresponding unit cells. t₀ is the time point at which we started increasing the temperature.

To elucidate the transformation mechanisms of these carbonates into their respective oxides, we tracked their evolution in real time. We first investigated the structural changes during the transformation of La₂(CO₃)₃·8H₂O as the temperature was increased from 23 to 800 °C under vacuum (Fig. 2A). With increasing temperature, the template gradually lost its crystallinity and became amorphous (Fig. 2A: t − t₀ = 0–3460 s), followed by recrystallization of this amorphous phase into a polycrystalline La₂O₃ phase (Fig. 2A: t − t₀ = 4660 s). The TGA profile of the template revealed two primary stages of weight loss (Fig. 2B): dehydrogenation occurring between 40 and 370 °C, and decarbonization occurring between 370 and 730 °C. These processes can be described by the following reactions:

La₂(CO₃)₃·8H₂O → La₂(CO₃)₃ + 8H₂O (1)

La₂(CO₃)₃ → La₂O₃ + 3CO₂ (2)

Fig. 2 Transformation pathway of the La₂(CO₃)₃·8H₂O template into La₂O₃. (A) TEM image series and the corresponding electron diffraction patterns of a La₂(CO₃)₃·8H₂O template transforming into La₂O₃ upon heating from 23 to 800 °C at a rate of 10 °C min⁻¹ in a vacuum. (B) TGA profile of La₂(CO₃)₃·8H₂O heated from 40 to 800 °C at a rate of 10 °C min⁻¹ in air. (C) Schematic diagram showing the transformation stages: La₂(CO₃)₃·8H₂O undergoes dehydrogenation into La₂(CO₃)₃ (step I), decarbonization into a-La₂O₃ (step II), and subsequent crystallization (step III). White arrows in (A) at t − t₀ = 1060–3460 s indicate the small voids that form within the template. The black arrow in (B) indicates the temperature at which the template composition approaches that of a stoichiometric La₂O₃. The dashed horizontal lines in (B) correspond to the theoretical weight percentages of La₂(CO₃)₃ and La₂O₃ derived from the decomposition of La₂(CO₃)₃·8H₂O. t₀ is the time point at which we started increasing the temperature.

Fig. 2C summarizes the transformation pathway. During the initial heating stage, most of the water molecules are removed from the template (I: dehydrogenation step), resulting in the loss of its crystallinity. In the subsequent stage, the CO₃²⁻ ions in La₂(CO₃)₃ decompose to form amorphous La₂O₃ (a-La₂O₃) and CO₂(g) (II: decarbonization step). Finally, the a-La₂O₃ undergoes crystallization into polycrystalline oxide (c-La₂O₃) (III: crystallization step). As shown in Fig. 2C, the interlayer and coordinated water molecules are released below 370 °C, indicating their minimal contribution to decarbonization and crystallization. In contrast, carbonate decomposition governs the evolution of La₂(CO₃)₃ into La₂O₃, thereby impacting its crystallinity, grain size, and defect density.^33,34 Note that the observed transformation pathway is independent of the heating rate; similar transformations were observed when La₂(CO₃)₃·8H₂O templates were heated at 25 and 100 °C min⁻¹ instead of 10 °C min⁻¹ (Fig. S2 and S3 vs.Fig. 2A).

To better visualize the crystallization of a-La₂O₃, we imaged the process at higher magnification, as shown in Fig. 3. When a-La₂O₃ was heated from 600 to 900 °C at a rate of 30 °C min⁻¹ (Fig. 3A), small La₂O₃ nuclei with sizes of 1–3 nm began to form at the edge of the template at 671 °C (Fig. 3B: t − t₀ = 142 s). Note that this temperature is significantly lower than the 730 °C required for the template composition to reach stoichiometric La₂O₃, as shown in Fig. 2B. This crystallization process is thermodynamically driven by the reduction in free energy,^35,36 while the plate-like template promotes heterogeneous nucleation by lowering the critical energy barrier through its large surface area and high-energy edge sites.³⁷ Upon further heating, these nuclei grew into larger grains and came into contact with adjacent grains, producing polycrystalline La₂O₃ (Fig. 3B: t − t₀ = 206–448 s).

Fig. 3 Nucleation of La₂O₃ nanocrystals in the amorphous film. (A) TEM and the corresponding FFT images showing the crystallization of the amorphous film into the polycrystalline La₂O₃ upon heating from 600 to 900 °C at a rate of 30 °C min⁻¹ in a vacuum (SI Video S1). (B) Sequence of inverse FFT images of the area (white box) in (A) showing the nucleation (t − t₀ = 130–142 s) and growth (t − t₀ = 206–448 s) of La₂O₃ nanocrystals (see Fig. S4 for the unprocessed TEM and FFT images). The amorphous and crystalline regions are false-colored in blue and yellow, respectively. t₀ is the time point at which we started increasing the temperature.

While La₂(CO₃)₃·8H₂O transforms into La₂O₃ without chemical interaction with the surrounding gaseous environment (eqn (1) and (2)), the thermal decomposition of other template materials can yield non-stoichiometric oxides. For example, in contrast to annealing in air, annealing Ce₂(CO₃)₃·8H₂O either under vacuum or N₂ produces oxygen-deficient CeO_2−x, in which cerium exists in both Ce⁴⁺ and Ce³⁺ oxidation states (Fig. S6 and S7). Hence, it is important to consider the effect of the atmosphere on the thermal decomposition process. To address this, we investigated the evolution of the Ce₂(CO₃)₃·8H₂O template during thermal decomposition under an oxygen atmosphere.

Our in situ TEM observations show that the transformation pathway of Ce₂(CO₃)₃·8H₂O closely resembles that of La₂(CO₃)₃·8H₂O, with both undergoing an amorphous intermediate phase before crystallizing into the respective metal oxides (Fig. 4A). The TGA profile of Ce₂(CO₃)₃·8H₂O exhibited two distinct weight-loss steps during this process (Fig. 4B). First, Ce₂(CO₃)₃·8H₂O underwent dehydrogenation to form Ce₂(CO₃)₃ at 40–270 °C. This was then followed by oxidative decarbonization of Ce₂(CO₃)₃ into CeO₂ at 270–330 °C, described by the following reactions:¹²

Ce₂(CO₃)₃·8H₂O → Ce₂(CO₃)₃ + 8H₂O (3)

Ce₂(CO₃)₃ + ½O₂ → 2CeO₂ + 3CO₂ (4)

Fig. 4 Transformation pathway of the Ce₂(CO₃)₃·8H₂O template into CeO₂. (A) TEM image series and the corresponding electron diffraction patterns of the Ce₂(CO₃)₃·8H₂O template transforming into CeO₂ upon heating from 23 to 600 °C at a rate of 10 °C min⁻¹ in O₂ (760 Torr of 20% O₂ and 80% Ar). (B) TGA profile of Ce₂(CO₃)₃·8H₂O heated from 40 to 600 °C at a rate of 10 °C min⁻¹ in air. (C) Schematic diagram showing the transformation stages: Ce₂(CO₃)₃·8H₂O undergoes dehydrogenation into Ce₂(CO₃)₃ (step I), oxidation and decarbonization into a-CeO₂ (step II), and subsequent crystallization (step III). (D) Sequence of inverse FFT images showing the nucleation (t − t₀ = 60–126 s) and growth (t − t₀ = 207–873 s) of CeO₂ nanocrystals (see Fig. S8 for the unprocessed TEM and FFT images) upon heating from 300 to 600 °C at a rate of 30 °C min⁻¹ in O₂ (SI Video S2). The amorphous and crystalline regions are false-colored in brown and green, respectively. The black arrow in (B) indicates the temperature at which the template composition approaches that of a stoichiometric CeO₂. The dashed horizontal lines correspond to the theoretical weight percentages of Ce₂(CO₃)₃ and CeO₂ derived from the decomposition of Ce₂(CO₃)₃·8H₂O. t₀ is the time point at which we started increasing the temperature.

At 330 °C, the composition of the template closely approached stoichiometric CeO₂ (Fig. 4B), which enabled the nucleation of CeO₂ nanocrystals within the amorphous phase (Fig. 4D).

Comparison of the atmosphere-dependent structural evolution of Ce₂(CO₃)₃·8H₂O and La₂(CO₃)₃·8H₂O templates reveals that O₂ significantly accelerates the transformation of Ce₂(CO₃)₃·8H₂O, but not La₂(CO₃)₃·8H₂O. At 300 °C, the Ce₂(CO₃)₃·8H₂O template retained its crystalline structure under vacuum (Fig. S6A: t − t₀ = 1660 s), but lost crystallinity under an oxygen atmosphere (Fig. 4A: t − t₀ = 1662 s). This behavior is consistent with oxidative decarbonization of Ce₂(CO₃)₃·8H₂O occurring at a much lower temperature (270–330 °C) compared to the non-oxidative decarbonization of La₂(CO₃)₃ (370–730 °C) (Fig. 4B and 2B).

This difference can be attributed to the greater positive charge density of Ce⁴⁺, which has a higher charge (+4) and a smaller ionic radius (0.87 Å) than those of La³⁺ (+3, 1.03 Å) and Ce³⁺ (+3, 1.01 Å).³⁸ The higher polarizing power of Ce⁴⁺ induces a stronger distortion of the CO₃²⁻ electron cloud, weakening its C–O bonds and thereby facilitating decomposition at substantially lower temperatures than in the cases of La³⁺ and Ce³⁺.

To further support this explanation, we calculated the CO₂ desorption energy (ΔE) as a proxy for the decarbonization step during the transformation of metal carbonates into oxides. Specifically, ΔE was evaluated for La₂O₂CO₃, Ce₂O₂CO₃, and Ce₂O₃CO₃, corresponding to CO₂ release from oxycarbonate intermediates containing La³⁺, Ce³⁺, and Ce⁴⁺, respectively. For example, La₂O₂CO₃ decomposes as La₂O₂CO₃ → La₂O₃ + CO₂. The calculated ΔE for the Ce⁴⁺-containing intermediate (Ce₂O₃CO₃) is significantly lower (101.4 eV) than that for the La³⁺- and Ce³⁺-containing intermediates (223.3 eV for La₂O₂CO₃ and 193.9 eV for Ce₂O₂CO₃) (Fig. S9).

Note that this carbonate-to-oxide transformation approach can be extended to other rare-earth systems. For example, Fig. S10 shows that Y₂(CO₃)₃·2H₂O transforms into Y₂O₃ and Pr₂(CO₃)₃·8H₂O transforms into Pr₆O₁₁. Taken together, these results suggest that, for rare-earth carbonates whose thermal decomposition does not involve changes in the metal oxidation states, the gaseous atmosphere under which these carbonates are annealed has little to no impact on their transformation to metal oxides.^12,39,40 By contrast, when oxidation to higher valence states is required, the presence of O₂ facilitates oxidation and accelerates the overall conversion process, whereas the lack of oxygen results in oxygen-deficient oxides with coexisting mixed-valence metal states.

Conclusions

Our in situ TEM studies reveal that the transformation of metal carbonate templates into metal oxides proceeds through a sequence of three distinct steps: dehydrogenation of the carbonate, followed by decarbonization into an amorphous metal oxide, and finally crystallization of the oxide. These results underscore the significance of real-time tracking in revealing transient structural states during crystallization, thereby providing mechanistic insights that can guide the controlled synthesis and optimization of oxide materials for a wide range of applications.

Methods

Materials

The following reagents from Sigma-Aldrich Co. (St Louis, MO, USA) were used to synthesize the La₂(CO₃)₃·8H₂O and Ce₂(CO₃)₃·8H₂O templates: lanthanum(III) nitrate hexahydrate (La(NO₃)₃·6H₂O, cat. no. 203548), cerium(III) nitrate hexahydrate (Ce(NO₃)₃·6H₂O, cat. no. 238538), sodium bicarbonate (NaHCO₃, cat. no. S6297), and ethylene glycol (EG, cat. no. 324558). Yttrium(III) nitrate hexahydrate (Y(NO₃)₃·6H₂O, cat. no. Y820618, Shanghai Macklin Biochemical Co., Ltd., Shanghai, China) and praseodymium(III) nitrate hexahydrate (Pr(NO₃)₃·6H₂O, cat. no. P106054, Shanghai Aladdin Biochemical Technology Co., Ltd., Shanghai, China) were used to synthesize the Y₂(CO₃)₃·2H₂O and Pr₂(CO₃)₃·8H₂O templates, respectively. Deionized (DI) water with a resistivity of 18.2 MΩ cm was used to prepare all aqueous solutions in this study.

Synthesis of carbonate templates

The La₂(CO₃)₃·8H₂O templates were synthesized based on a modified procedure of Yin et al.¹² Specifically, 0.23 g of NaHCO₃, 2 mL of EG, and 7 mL of DI water were mixed in a 50 mL centrifuge tube. Next, 0.44 g of La(NO₃)₃·6H₂O was dissolved in 3 mL of DI water, and the solution was added to the centrifuge tube. The resulting mixed solution was maintained at ambient temperature for 6 h. The precipitate was collected by centrifugation, washed five times with DI water, and dried overnight at 60 °C.

The Ce₂(CO₃)₃·8H₂O, Y₂(CO₃)₃·2H₂O, and Pr₂(CO₃)₃·8H₂O templates were synthesized following the same procedures as the La₂(CO₃)₃·8H₂O templates, but instead of La(NO₃)₃·6H₂O, 0.44 g of Ce(NO₃)₃·6H₂O, Y(NO₃)₃·6H₂O, or Pr(NO₃)₃·6H₂O was used, respectively. Additionally, for Ce₂(CO₃)₃·8H₂O, the template solution was maintained at ambient temperature for 3 h rather than 6 h.

In situ TEM experiments

A Titan S/TEM (Thermo Fisher Scientific Ltd., Hillsboro, OR, USA) operated at an accelerating voltage of 300 kV was used for in situ TEM studies. Image series were acquired at 5 frames per second using a Gatan K2 IS camera (Gatan Inc., Pleasanton, CA, USA), with an electron flux ranging from 100 to 200 e⁻ Å⁻² s⁻¹.

DENSsolutions Wildfire and Climate TEM holders (DENSsolutions, Delft, Netherlands) were used for in situ heating studies of the templates in a vacuum and in a gaseous environment (20% O₂ and 80% Ar at 1 atm pressure), respectively. The gas mixture (O₂ + Ar) was flowed into the TEM holder at a flow rate of 40–50 μL min⁻¹ using the DENSsolutions gas delivery system.

Schematics of crystal structures

The crystal structures of the carbonate templates and the crystallized oxides shown in Fig. 1A and B were adapted from ref. 12, 30–32 and rendered using the CrystalMaker® 9.0.3 software package.⁴¹ The LaO₁₀, CeO₁₂ and CO₃ polyhedra are depicted in translucent blue, green, and grey, respectively.

Electron diffraction profiles

The radial diffraction profiles in Fig. S6B are the plots of the summed intensity S(k) obtained from electron diffraction images after background subtraction. The summed intensity was calculated using the following equation:

S(k) = ∑I(k) (5)

Here, k = (k_x,k_y) and are the outward radial vector from the center of the diffraction pattern and its corresponding length, and I(k) is the diffraction intensity at that point.

Thermal decomposition in a tube furnace

Thermal decomposition of the templates was carried out in a tube furnace (NBD-O1200-50IT-80F, Henan Nobody Materials Science and Technology Co., Zhengzhou, China) under either air or an N₂ environment. For each study, 100–200 mg of a template was placed in a quartz boat, which was then positioned at the center of a quartz tube with an inner diameter of 42 mm. For decomposition in air, the samples were heated from room temperature to 800 °C at a rate of 10 °C min⁻¹ under static air. For decomposition under N₂, the tube was pumped down and purged with pure N₂ three times prior to heating. The samples were then heated from room temperature to 800 °C at the same rate under a continuous N₂ flow of approximately 200 sccm, sufficient to create a slight overpressure relative to the ambient atmosphere.

XRD, TGA, and XPS characterization

The XRD patterns shown in Fig. S1 and S5 were obtained using a Bruker D8 Advance Powder X-ray Diffractometer (Bruker, Billerica, MA, USA) with a 2θ range of 5°–70° and a step size of 0.02° s⁻¹. The XRD patterns shown in Fig. S10 were obtained using a PANalytical X'Pert PRO Powder X-ray Diffractometer (Malvern Panalytical, Almelo, The Netherlands) with a 2θ range of 5°–80° and a step size of 0.02° s⁻¹.

The TGA profiles displayed in Fig. 2B and 4B were acquired using a Discovery TGA (TA Instruments, New Castle, DE, USA). During TGA measurements, the samples were heated in air from 40 to 600 or 800 °C at a rate of 10 °C min⁻¹.

The XPS results shown in Fig. S7 were obtained using a Kratos Axis Ultra DLD spectrometer (Kratos Analytical, Shimadzu, Japan), equipped with a monochromatized Al Kα X-ray source (15 kV, 3 mA, beam size 300 × 700 μm²). The pass energy was set to 40 eV for the high-resolution scans. Raw XPS data were calibrated using the binding energy of the C 1s peak at 284.8 eV, and background subtraction was performed using the Shirley method.⁴² Data analysis was performed with XPSPEAK 4.1 software.⁴³ The Ce 3d spectra were deconvoluted into Ce³⁺ and Ce⁴⁺ components. The fractions of Ce³⁺ (f_Ce³⁺) and Ce⁴⁺ (f_Ce⁴⁺) shown in Fig. S7A were calculated using the following equations:⁴⁴

A_Ce³⁺ = A_v′ + A_u′ (6)

A_Ce⁴⁺ = A_v + A_v′′ + A_v′′′ + A_u + A_u′′ + A_u′′′ (7)

Here, A_v′ and A_u′ represent the deconvoluted peak areas at binding energies of 885.2 and 903.1 eV, respectively, corresponding to Ce³⁺ contribution. The Ce⁴⁺ contribution is given by the peak areas A_v, A_v′′, A_v′′′, A_u, A_u′′, and A_u′′′ associated with binding energies of 882.4 (v), 888.8 (v′′), 898.0 (v′′′), 900.7 (u), 907.3 (u′′), and 916.6 eV (u′′′), respectively.

Computational methods

All calculations displayed in Fig. S9 were performed using the Gaussian 09 software package.⁴⁵ The B3LYP hybrid density functional theory (DFT) method^46,47 was used in combination with the CEP-31G basis set,^48–50 which includes effective core potentials for rare-earth elements such as cerium and lanthanum. The charge and spin multiplicity were assigned according to the expected oxidation states of the metal centers. Single-point energy calculations were performed on optimized geometries of nanoscale clusters to refine the electronic energy values. This computational setup provides a reliable balance between accuracy and efficiency for systems containing f-block elements.⁵¹

The geometric structures of oxide clusters (La₂O₃, Ce₂O₃, and CeO₂), their corresponding oxycarbonate intermediates (La₂O₂CO₃, Ce₂O₂CO₃, and Ce₂O₃CO₃), and CO₂ were fully optimized without symmetry constraints. Vibrational frequency analyses were performed, and the absence of imaginary frequencies confirmed that these structures correspond to optimized geometries located at stable local minima on the potential energy surface, rather than transition states or saddle points.

The CO₂ desorption energy (ΔE) from oxycarbonate intermediates was calculated as a proxy for the decarbonization step of metal carbonates using:

ΔE = E_oxide + E_CO2 − E_oxycarbonate (10)

Here, E_oxide, E_CO2, and E_oxycarbonate are the total energies of the isolated oxide cluster, the CO₂ molecule, and the oxycarbonate intermediate, respectively.

Author contributions

Y. J. and U. M. conceived the study. Y. J. conducted the experiments and analyzed the data. G. L. performed the DFT simulations. Y. J. and U. M. wrote the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5nr01525a.

Acknowledgements

This work was supported by the Singapore National Research Foundation's Competitive Research Program funding (NRF-CRP23-2019-0001) and the Ministry of Education-Singapore (MOE-T2EP10220-0003).

References

1. T. Montini, M. Melchionna, M. Monai and P. Fornasiero, Chem. Rev., 2016, 116, 5987–6041.
2. T. Mori, J. Drennan, J.-H. Lee, J.-G. Li and T. Ikegami, Solid State Ionics, 2002, 154–155, 461–466.
3. K. Xiong and J. Robertson, Appl. Phys. Lett., 2009, 95, 022903.
4. S. B. Brachetti-Sibaja, S. E. Rodil, M. A. Domínguez-Crespo, A. M. Torres-Huerta, E. Rodríguez and A. B. López-Oyama, Thin Solid Films, 2017, 636, 615–621.
5. A. Wang, W. Quan, H. Zhang, H. Li and S. Yang, RSC Adv., 2021, 11, 20465–20478.
6. H. Wang, W. Lin, X. Qiu, F. Fu, R. Zhong, W. Liu and D. Yang, ACS Sustainable Chem. Eng., 2018, 6, 3696–3705.
7. R. Prins, J. Catal., 2020, 392, 336–346.
8. S. K. Meher and G. R. Rao, J. Colloid Interface Sci., 2012, 373, 46–56.
9. L. Xu, Z. Li, Q. Cai, H. Wang, H. Gao, W. Lv and J. Liu, CrystEngComm, 2010, 12, 2166–2172.
10. L. Zhu, S. Pu, K. Liu, T. Zhu, F. Lu and J. Li, Mater. Lett., 2012, 83, 73–75.
11. T. Ye, W. Huang, L. Zeng, M. Li and J. Shi, Appl. Catal., B, 2017, 210, 141–148.
12. S. Yin, Y. Minamidate, S. Tonouchi, T. Goto, Q. Dong, H. Yamane and T. Sato, RSC Adv., 2012, 2, 5976–5982.
13. I. Bretos, R. Jiménez, J. Ricote and M. L. Calzada, Chem. Soc. Rev., 2018, 47, 291–308.
14. Q. Mu and Y. Wang, J. Alloys Compd., 2011, 509, 396–401.
15. R. Kavitha, S. R. Hegde and V. Jayaram, Mater. Sci. Eng., A, 2003, 359, 18–23.
16. M. Hu, W. Yang, H. Tan, L. Jin, L. Zhang, P. Kerns, Y. Dang, S. Dissanayake, S. Schaefer and B. Liu, et al. , Matter, 2020, 2, 1244–1259.
17. N. W. Kwak and W. Jung, Acta Mater., 2016, 108, 271–278.
18. E. A. Villegas, C. M. Aldao, R. Savu, L. A. Ramajo and R. Parra, Phys. Status Solidi A, 2018, 215, 1800107.
19. T. Arii, A. Kishi, M. Ogawa and Y. Sawada, Anal. Sci., 2001, 17, 875–880.
20. R. Warringham, S. Mitchell, R. Murty, R. Schäublin, P. Crivelli, J. Kenvin and J. Pérez-Ramírez, Chem. Mater., 2017, 29, 4052–4062.
21. C. Vaysse, L. Guerlou-Demourgues and C. Delmas, Inorg. Chem., 2002, 41, 6905–6913.
22. J. Pérez-Ramírez, G. Mul, F. Kapteijn and J. A. Moulijn, J. Mater. Chem., 2001, 11, 821–830.
23. J. Pérez-Ramırez, G. Mul and J. A. Moulijn, Vib. Spectrosc., 2001, 27, 75–88.
24. Y. Zhao, X. Jia, G. Chen, L. Shang, G. I. Waterhouse, L.-Z. Wu, C.-H. Tung, D. O'Hare and T. Zhang, J. Am. Chem. Soc., 2016, 138, 6517–6524.
25. Y. Jiang, M. Duchamp, S. J. Ang, H. Yan, T. L. Tan and U. Mirsaidov, Nat. Commun., 2023, 14, 104.
26. Y. Jiang, Z. M. Wong, H. Yan, T. L. Tan and U. Mirsaidov, ACS Nano, 2025, 19, 3886–3894.
27. T. Ghosh, J. M. Arce-Ramos, W.-Q. Li, H. Yan, S. W. Chee, A. Genest and U. Mirsaidov, Nat. Commun., 2022, 13, 6176.
28. S. W. Chee, J. M. Arce-Ramos, W. Li, A. Genest and U. Mirsaidov, Nat. Commun., 2020, 11, 2133.
29. Y. Jiang, A. M. H. Lim, H. Yan, H. C. Zeng and U. Mirsaidov, Adv. Sci., 2023, 10, 2302663.
30. D. B. Shinn and H. A. Eick, Inorg. Chem., 1968, 7, 1340–1345.
31. R. Sarbajna, A. S. Devi, K. Purandhar and M. V. Suryanarayana, Int. J. ChemTech Res., 2013, 5, 2810–2820.
32. S. N. Kabekkodu, A. Dosen and T. N. Blanton, Powder Diffr., 2024, 39, 47–59.
33. L. Zeng, D. B. Buchholz, D. T. Keane, T. J. Marks, J. E. Medvedeva and M. J. Bedzyk, Chem. Mater., 2024, 36, 5965–5975.
34. Y. Wang, H. Wang, C. S. Li, J. Q. Feng, L. H. Jin, Z. M. Yu, P. Odier and P. X. Zhang, Ceram. Int., 2015, 41, 14270–14275.
35. C. Yang, Y. Hirose, S. Nakao, N. L. H. Hoang and T. Hasegawa, Appl. Phys. Lett., 2012, 101, 052101.
36. J. L. M. Rupp, C. Solenthaler, P. Gasser, U. P. Muecke and L. J. Gauckler, Acta Mater., 2007, 55, 3505–3512.
37. N. Karpukhina, R. G. Hill and R. V. Law, Chem. Soc. Rev., 2014, 43, 2174–2186.
38. R. D. Shannon, Acta Crystallogr., Sect. A, 1976, 32, 751–767.
39. L. Paama, I. Pitkänen, H. Halttunen and P. Perämäki, Thermochim. Acta, 2003, 403, 197–206.
40. R. P. Turcotte, J. O. Sawyer and L. Eyring, Inorg. Chem., 1969, 8, 238–246.
41. CrystalMaker 9.0.3, CrystalMaker Software Ltd, Oxford, England, 2014, https://www.crystalmaker.com.
42. J. Végh, J. Electron Spectrosc. Relat. Phenom., 2006, 151, 159–164.
43. R. W. M. Kwok, XPSPEAK 4.1, The Chinese University of Hong Kong, Hong Kong, 2000.
44. C. Anandan and P. Bera, Appl. Surf. Sci., 2013, 283, 297–303.
45. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, et al., Gaussian 09, Revision A.02, Gaussian Inc., Wallingford CT, US, 2016.
46. A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
47. C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785–789.
48. W. J. Stevens, H. Basch and M. Krauss, J. Chem. Phys., 1984, 81, 6026–6033.
49. W. J. Stevens, M. Krauss, H. Basch and P. G. Jasien, Can. J. Chem., 1992, 70, 612–630.
50. T. R. Cundari and W. J. Stevens, J. Chem. Phys., 1993, 98, 5555–5565.
51. T. Miyazaki, S. Okita, T. Ohta, H. Yagi, R. Sumii, H. Okimoto, Y. Ito, H. Shinohara and S. Hino, Chem. Phys., 2015, 447, 71–75.

---