Core-shell nanostructures have attracted significant interest due to their charming configurations and intercomponent synergistic effects that might enhance their expected functions or launch exotic properties [1–21]. These have rendered the core-shell nano-structures one of prime targets in the development of advanced composite materials. Of them, intensive synthetic efforts have been invested into the development of inorganic core-shell metal nanoparticles during the past decade [1–13]. Precise construction of core-shell species with well-defined molecular structures and compositions is of significance to atomic-level understanding and regulation of their properties. However, although some effective approaches have been explored to precisely control the size, shape, thickness and composition of core-shell metal nanoparticles, it is still an enormous challenge to achieve their atomic-level bonding geometry and structural characterization.

Crystalline inorganic core-shell metal oxide nanoclusters with atomically and geometrically precise structural information are of great interest not only due to their fascinating structural features with unusual physical and chemical properties [22–29], but also important roles as the molecular models for the structures and physicochemical properties of inorganic core-shell metal oxide nanoparticles. Particularly, precisely tuning material compositions at the atomic scale remains an attractive target and scientifically interesting not only for developing novel composite materials with controllable composition but also providing atomic-level insights into the structures and physicochemical properties of amorphous core-shell heterometal nanoparticles.

Nevertheless, the targeted growth of inorganic core-shell heterometal oxide nanoclusters with atomic precision is still beyond our capacity owing to the difficulty of atomic-level design of both core and shell and the complexity of self-assembly reactions. And thus, the development of such species is still being sought, lagged far behind that of inorganic core-shell metal nanoparticles. At present, known such nanoclusters are limited and some representative examples are Cu₂₀@W₄₈ [24], V₆@W₄₈ [25], M₁₆@W₄₈ (M = Fe [26], Al or Ga [27]), Bi₆@U₂₄ and Pb₈@U₂₄ [28].

Recently, we reported the synthesis of an all-inorganic core-shell-type cobalt polyoxoniobate (PONb) Na₈K₁₄H₁₀[Co₂₆@Nb₃₆]·nH₂O (1, Co₂₆@Nb₃₆ = [Co₂₆Nb₃₆O₁₄₀]³²⁻), in which a large 36-nuclearity PONb shell {Nb₃₆O₁₂₄} (Nb₃₆) encloses a 26-nuclearity mixed-valence cobalt oxide core {Co₂₆O₃₂} (Co₂₆) (Fig. 1) [30]. Interestingly, the core composition of Co₂₆@Nb₃₆ is sensitive to the reaction conditions, allowing to precisely tune its inner Co-core content without affecting its outer Nb-shell composition and structure, resulting in a series of isostructures {Co_26-nNb_36+nO₁₄₀} (n = 0–2). Proton conduction experiments reveal that these materials have good proton conducting properties.

Solid 1 crystallized in a highly symmetric cubic space group Fm-3m. The 36-nuclearity Nb₃₆ shell structure can be described as a nanosized T_d-symmetry tetrahedral nanocage comprising four {Nb₃O₁₃} triangles positioned at its four vertices and four {Nb₆O₂₄} hexagonal rings laid flat on its four faces (Fig. 2a), which are interlinked with each other via corner-sharing. The 26-nuclearity Co₂₆ core, also with T_d-symmetry, contains a tetrahedral {Co^Ⅲ₁₆O₂₀} core and ten Co^ⅡO₄ tetrahedra. The {Co^Ⅲ₁₆O₂₀} core can be seen as consisting of five Co^Ⅲ₄O₄ cubanes, wherein the innermost Co^Ⅲ₄O₄ cubane joins four tetrahedrally arranged Co^Ⅲ₄O₄ cubanes by sharing four Co^Ⅲ atoms (Fig. 2b). Further, the ten Co^ⅡO₄ tetrahedra occupy the six edge positions (in blue) and the four face-centered positions (in yellow) of the tetrahedral {Co^Ⅲ₁₆O₂₀} core via corner-sharing with Co^ⅢO₆ octahedra, giving the overall mixed-valence Co₂₆ cluster.

Interestingly, the core composition of 1 is markedly affected by the reactants, allowing us to precisely manipulate its Co-core composition with atomic-level precision while keeping the same Nb-shell composition and topology. To begin with, we found that the starting phosphates of Na₃PO₄ and KH₂PO₄ in the reaction of 1 play a key role in determining the resulting core-shell nanocluster with an ordered or disordered core structure though they do not present in the products. Specifically, the reaction of 1 without the addition of phosphates yields an isomorph Na₁₀K₁₅H₅[NbCo₂₅@Nb₃₆]·nH₂O (2) (NbCo₂₅@Nb₃₆ = [Co₂₅Nb₃₇O₁₄₀]³⁰⁻). As shown in Fig. 3, the difference between Co₂₆@Nb₃₆ and NbCo₂₅@Nb₃₆ lies in that the four equivalent Co³⁺ sites (related by a S₄ axis) of the innermost Co^Ⅲ₄O₄ cubane in the former are found to be occupied either by Nb⁵⁺ or Co³⁺ ions in the latter. The disordered site is refined freely to give a Co/Nb occupancy factor of 0.758/0.242, followed by fixing it to 0.75/0.25, and thus the core composition of Co₂₅Nb@Nb₃₆ is NbCo₂₅O₃₂.

The observed Nb/Co disorder in Co₂₅Nb@Nb₃₆ reveals that Nb and Co ions compete to form the innermost cubane unit in the solution, and the addition of Na₃PO₄ and KH₂PO₄ favors the formation of ordered Co₂₆@Nb₃₆. Attempts to obtain analogs of Co₂₆@Nb₃₆ proved fruitless by the replacment of the phosphates with nitrate, carbonate, sulfate, acetate or silicate. Considering that core-shell structures are usually grown from inside out, the distinct innermost cubanes of Co₄O₄ and NbCo₃O₄ indicate that the starting phosphates Na₃PO₄ and KH₂PO₄ should have a unique influence on the initial nucleation of the core-shell nanocluster. Unfortunately, the detailed role of the phosphates still remains unclear. Potentially, they can affect the competitive relationship between Nb⁵⁺ and Co³⁺ ions for the initial nucleation of core-shell nanoclusters via hydrogen bond interactions or affect the pH value of the reaction solution and thus the reaction and crystallization processes.

Given that Nb and Co ions can compete to form the innermost metal cubane, we set out to explore the molecular growth of the core-shell Co-Nb clusters by adjusting the starting Co/Nb ratio to see if it can affect the core composition of resulting cluster. We studied the Co/Nb ratio influence over the range from 1:9 to 1:1, with the amount of K₇HNb₆O₁₉·13H₂O fixed at 0.33 mmol. When the Co/Nb ratios are in the range of 1:9~1:5 (noted that the Co/Nb ratio in the reaction of 2 is about 1:8), the obtained core-shell nanoclusters are all the same as NbCo₂₅@Nb₃₆ found in 2, as confirmed by single-crystal X-ray diffraction analyses of four single crystals from the four same reactions as 2 but with different initial Co/Nb ratios of 1:9, 1:7, 1:6 and 1:5, respectively. Increasing the Co/Nb ratio to 1:4 results in the formation of a new isostructural compound Na₁₀K₁₅H₃[Nb₂Co₂₄@Nb₃₆]·nH₂O (3) (Nb₂Co₂₄@Nb₃₆ = [Co₂₄Nb₃₈O₁₄₀]²⁸⁻), in which all trivalent Co³⁺ sites, that is, all Co sites of the whole tetrahedral {Co^Ⅲ₁₆O₂₀} core in Co₂₆@Nb₃₆, are found as disordered Co/Nb sites in Nb₂Co₂₄@Nb₃₆ (Fig. 3). The asymmetric unit of the {M₁₆O₂₀} (M stands for disordered Co/Nb sites) core in Nb₂Co₂₄@Nb₃₆ contains two crystallographic independent M sites of M1 and M2, which are refined freely to give Co/Nb occupancy factors of 0.789/0.211 and 0.892/0.108, respectively, followed by fixing them to 0.8/0.2 and 0.9/0.1. Since the {M₁₆O₂₀} consists of four M1 and twelve M2, its composition exactly is {Nb₂Co₁₄}. Similar to the bivalent Co²⁺ sites in Co₂₆@Nb₃₆ and NbCo₂₅@Nb₃₆, all Co²⁺ sites in Nb₂Co₂₄@Nb₃₆ are found to be full occupancy. And thus, the overall core composition is Nb₂Co₂₄. We further increased the Co/Nb ratios from 1:3 to 1:1 with attempt to get new isostructres and to check if these tetrahedrally coordinated Co²⁺ sites can be disordered as Co/Nb sites, however, no isostructural species but amorphous phases were obtained.

The distinct inner core compositions make NbCo₂₅@Nb₃₆ and Nb₂Co₂₄@Nb₃₆ nanoclusters have different cluster charges of −30 and −28, respectively. Considering that the only difference in the syntheses of them is the starting amount of CoCl₂, we attribute the formation of the distinct isostructures NbCo₂₅@Nb₃₆ and Nb₂Co₂₄@Nb₃₆ to the role of the Co²⁺ ions on the mediation of solution pH via hydrolysis. It is known that, in basic solution, a higher solution pH will stabilize a higher-charge polyoxoanion. With different starting Co/Nb ratios from 1:9 to 1:5, the pH values after reactions are about 12.5 ± 0.1 (Table S1 in Supporting information), suggesting the Co/Nb ratios in the range from 1:9 to 1:5 have little effect on the pH values of reactions. Accordingly, all these reactions yield the same nanocluster of NbCo₂₅@Nb₃₆ with a net charge of −30. While, when the Co/Nb ratio increases to 1:4, the pH value after reaction decreases to 11.9, suggesting that this ratio has an obvious effect on the pH value of reaction. Accordingly, the reaction gives Nb₂Co₂₄@Nb₃₆ with a lower net charge (−28) compared to NbCo₂₅@Nb₃₆. We can therefore speculate that the hydrolysis of different Co²⁺ concentrations lead to reaction solutions with different pH values, which make in-situ generating intermediates adjust their inner core compositions to form isostructural clusters with different negative charges that could be adapted to different pH values of reaction solutions.

The above results reveal that the atomic-level modulation of core-shell Co₂₆@Nb₃₆ cluster with order/disorder core structure and programmable core composition can be achieved by controlling the starting reagents. The obtainment of the series of core-shell Co-Nb bimetallic clusters constitutes a family with a general formula of {Co_26-nNb_36+nO₁₄₀} (n = 0–2 in this work). These nanoclusters might provide atomic-level insights into the influence of the starting chemicals on the structures and compositions of amorphous bimetallic core-shell nanoparticles. Additionally, the observed disordered core structure evolution from {Co₄O₄} to {NbCo₃O₄}, and to {Nb₂Co₁₄O₂₀} provides a case to reveal an inside-out molecular growth of core-shell bimetal nanoclusters at atomic level.

The metal contents of 2–3 were determined by ICP analyses, which are in good agreement with the calculated values based on single-crystal X-ray diffraction analyses (Table S2 in Supporting information). Further, XPS were performed to confirm their ratios in 2–3 (Fig. S1 in supporting information). Curve-fitting of the high-resolution Co 2p spectra reveals that the Co³⁺/Co²⁺ ratios in 2–3 are 1.48 and 1.38, respectively, perfectly matching their corresponding Co³⁺/Co²⁺ ratios of 1.50 and 1.40 identified by the BVS calculations [31].

High chemical stabilities, the presence of counter cations (H⁺, Na⁺, and K⁺) and oxo-rich surface point to the possible utility of 1 as proton conducting material. The conductivity of 1 was investigated by ac impedance measurements using a two-electrode configuration between 10⁷ and 1 Hz. The humidity-dependent conductivity was first measured at room temperature (RT) (25 ℃). As dipicted in Fig. 4a, the conductivity is 2.68 × 10⁻⁴ S/cm at 55% relative humidity (RH), and it incerases to 1.74 × 10⁻³ S/cm when RH maintianed at 98%. Next, the temperature-dependent proton conductivity was measured over the temperature range of 25–85 ℃ under 98% RH. The result reveal that the conductivity of 1 reaches 2.64 × 10⁻² S/cm at 85 ℃ (Fig. 4b), which is among the highest conductivity reported for POMs (Table S3 in Supporting information). According to the Arrhenius equation σT = σ₀exp(-E_a/k_BT), the activation energy, determined by linear regression analysis, is estimated to 0.304 eV (Fig. 4c), indicating that conduction is mainly carried out by the "Grotthus" mechanism (E_a < 0.4). We attribute the excellent proton conductivity to the presence of rich proton carriers within the structure of 1, including water molecules and terminal oxygen atoms, which help to form hydrogen-bonded proton "hopping" networks [32–35]. Additionally, the PXRD patterns showed that the sample remain consistent before and after conduction tests, revealing the integrity of sample 1 (Fig. 4d).

Although extensive research efforts have been devoted to studying the proton-conducting performances of POMs [36–40], most of them are based on compacted crystalline powder samples. The research on the conductivities of single-crystal POMs remains largely unexplored. Given that 1–3 contain different amounts of cations and might exhibit distinct conductivities, the single crystal proton conductivities of 1–3 were further investigated by ac impedance measurements using single crystals over the temperature range of 35–75 ℃ under 98% RH. Ag wires were carefully attached to a cubic crystal (0.20 mm × 0.20 mm × 0.20 mm) by using conductive silver paste along opposite faces. The temperature-dependent conductivity measurements reveal that the singel crystal conductivity of 1 at 35 ℃ is 2.2 × 10⁻⁶ S/cm, which is comparable to those of some reported single-crystal POMs [39,40]. As the temperature increases, the conductivity increases gradually and reaches a maximum of 1.08 × 10⁻⁵ S/cm at 75 ℃ (Fig. 5a). The conductivities of 2 and 3 at 35 ℃, 75 ℃ with 98% RH are 1.54 × 10⁻⁶ S/cm, 6.84 × 10⁻⁶ S/cm and 1.92 × 10⁻⁶ S/cm, 5.89 × 10⁻⁶ S/cm, respectively (Figs. 5b and c). The results show that the conductivities of 1–3 are almost at the same level. Nevertheless, crystal 1 with the most cations shows relatively high conductivity compared to crystals 2 and 3. According to the Arrhenius equation σT = σ₀exp(-E_a/k_BT), the activation energy of single crystals 1–3, determined by linear regression analysis, are 0.354, 0.346 and 0.260 eV, respectively, indicating that their conduction processes are mainly dominated by the "Grotthus" mechanism (Fig. 5d).

In summary, we demonstrate that the inner core structure (in order/disorder) and composition (with different Co/Nb distributions) of a unique all-inorganic core-shell-type Co-Nb heteropolyoxoniobate Co₂₆@Nb₃₆ can be precisely modulated by simple control of initial reaction reagents, leading to a family of isostructural core-shell nanoclusters {Co_26-nNb_36+nO₁₄₀} (n = 0–2) with precise atoms and structures. Further, their proton conductivities have been examined to showcase them as promising candidates for proton-conducting materials. These bimetallic core-shell structures with tunable core structures allow studies in rich host-guest chemistry to be envisaged. Furthermore, we anticipate that other novel transition metal clusters can be captured, resulting in new discoveries in metal oxo cluster chemistry. Finally, this series of rare all-inorganic and atomically precise Co-Nb core-shell nanoclusters might provide a foundation to understand and investigate the chemical processes of Co-Nb-based non-molecular counterparts including inorganic core-shell nanoparticles and bulk materials.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We gratefully acknowledge the financial support from the National Natural Science Foundation of China (Nos. 21971039 and 22171045) and the Key Program of Natural Science Foundation of Fujian Province (No. 2021J02007).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.109124.