Introduction

The programmable construction of metal-organic polyhedra (MOPs) represents a central goal in supramolecular chemistry1,2,3,4,5,6,7, driven by their potential for creating tailored nanocavities with applications in separation, catalysis, and sensing8,9,10,11,12. A significant barrier to achieving this goal is the inherent difficulty in probing self-assembly mechanisms, which typically involve high-energy intermediates13. While in situ techniques such as nuclear magnetic resonance (NMR)14,15, high-resolution mass spectrometry (HRMS)16, and fluorescence spectrum17 can track coordination processes in real time to some extent, they often lack the structural resolution needed to unambiguously identify intermediate species.

This challenge is particularly acute in titanium chemistry, where the high reactivity and oxophilicity of Ti4+ ions18,19,20 typically lead to the formation of thermodynamically stable titanium-oxo clusters (TOCs)21,22,23,24,25 or metal-organic frameworks (Ti-MOFs)26,27,28,29,30, rather than discrete Ti-MOPs31,32,33,34. Since Raymond’s pioneering report on tetrahedral Ti4 MOP in 199835, although certain progress has been made in this field, there are still fewer than 20 types of Ti-MOPs to date31. These structures still have the following limitations: i) low nuclearities (≤18 Ti atoms)34, ii) simplistic topologies such as tetrahedra and cubes36,37,38 (Fig. 1a), and iii) low modifiability. The lack of structural complexity and functional versatility in known Ti-MOPs highlights a significant gap in the field. Although our group has contributed to this area with the synthesis of a giant cage-like Ti42 oxo-cluster and several Ti-MOPs36,39,40,41, the development of high-nuclearity Ti-MOPs with sophisticated architectures remains a formidable challenge, owing to the difficulty in controlling Ti4+ ion reactivity and hydrolysis.

Fig. 1: Structural evolution and hierarchical assembly of Ti-MOPs.
Fig. 1: Structural evolution and hierarchical assembly of Ti-MOPs.
Full size image

a Crystal structures and publication years of representative Ti-MOPs: Ti4 (ref. 35), Ti8 (ref. 38), Ti12 (ref. 32), Ti14 (ref. 33), Ti18 (ref. 34), and Ti24 (FIR-151, this work). b Proposed hierarchical assembly mechanism of the Ti24 MOP, illustrating the stepwise formation from primary building units to the final polyhedral architecture. Atom colors: Ti, sky blue; C, gray; O, red. All hydrogen atoms are omitted for clarity.

Herein, we have successfully synthesized and determined the crystal structure of FIR-151—a Ti24 MOP with tro topology. This high-symmetry cage not only represents the highest nuclearity Ti-MOP reported to date, but also demonstrates how the inherent challenges of titanium coordination chemistry can be overcome to achieve architectural complexity. The assembly pathway was elucidated through the structural characterization of two key intermediates: a Ti12 macrocycle (FIR-152) and a Ti12 (6-4-6) module (FIR-153). These structures provide direct evidence for a hierarchical assembly pathway (Fig. 1b), underpinned by a coordination lability gradient between Ti4+ and Ni2+. Furthermore, FIR-151 exhibits permanent porosity, selective gas adsorption, and post-assembly modification capabilities, demonstrating functional relevance. This work not only advances the understanding of titanium supramolecular chemistry but also establishes an effective strategy for achieving programmable synthesis in dynamic metal-organic systems.

Results

Squaric acid (H2C4O4) is frequently employed as an organic template to facilitate the ring expansion and assembly of macrocyclic compounds42. In existing titanium complexes, it has been observed that the C4O42– ligand coordinates with four Ti4+ ions to form a 4-membered ring module43,44. Such modules are often found in complex polyhedral structures, exhibiting topological types such as sod, rho, and lta. Here, we select Ti4+ ions and C4O42– ligands for assembly to construct polyhedral featuring intricate topological architectures. The reaction of Ti(OiPr)4 with H2C4O4 in the mixture solvent of acetic acid (CH3COOH)/acetonitrile (CH3CN) at 120 °C led to the formation of yellow rhombic dodecahedral crystals of FIR-151 (FIR = Fujian Institute Research; Supplementary Fig. 1). Single-crystal X-ray diffraction (SCXRD) analysis reveals that FIR-151 crystallizes in the tetragonal P42/nmc space group with the formula as Ti24O24(C4O4)6(CH3COO)36·guest (Supplementary Fig. 2 and Table S1). Its structure features a Ti24 MOP with a tro topology, representing the highest nuclearity Ti-MOP reported to date (Fig. 2a).

Fig. 2: Structural and hierarchical assembly of the Ti24 MOP and its assembly intermediates.
Fig. 2: Structural and hierarchical assembly of the Ti24 MOP and its assembly intermediates.
Full size image

a Molecular structure of FIR-151. b Topological representation of FIR-151, with spheres denoting Ti atoms (vertices) and rods representing Ti-Ti edges. The Ti12 (6-4-6) module is highlighted in red. c Graphical representations of the truncated octahedron and its corresponding net, composed of six squares and eight hexagons. Molecular structures of the intermediate species and kinetic trapper: the Ti12 macrocycle (intermediate 1, d the Ni3Cl4 cluster (kinetic trapper, e and Ti12 (6-4-6) modular (intermediate 2, f). g Proposed hierarchical assembly mechanism of the truncated octahedral-type MOP. Atom colors: Ti, sky blue; Ni, purple; C, gray; O, red; Cl, green; N, blue. All hydrogen atoms are omitted for clarity.

To elucidate the structure of FIR-151, we conducted a detailed analysis (Supplementary Fig. 3). Firstly, twelve Ti atoms are bridged by twelve μ2-O atoms to form a Ti12 macrocycle—marking what is, to the best of our knowledge, the first observation of such a Ti12 structure (Supplementary Table S2). This ring is capped by eighteen terminally coordinated CH3COO– ligands. Then, one C4O42– ligand positions within the central pore of the Ti12 ring, coordinating to four Ti atoms, leading to the formation of a Ti12 (6-4-6) module. Four additional C4O42– ligands occupy the remaining coordination sites on the Ti atoms within this module. The incorporation of C4O42– induces a conformational change in the Ti12 macrocycle, causing it to narrow and bend into a distinct roof-like architecture. This contrasts sharply with the planar or S-shaped configurations typically found in other wheel-type TOCs (Supplementary Fig. 4)25. The cage structure of FIR-151 can be viewed as a dimer of two Ti12 (6-4-6) modules via four shared C4O42– ligands (Fig. 2b and Supplementary Fig. 5). The assembly pathway of FIR-151, involving the dimerization of two curved Ti12 (6-4-6) modules, is reminiscent of the formation of iconic supramolecular structures such as the ‘tennis ball’ capsule45. In that system, two identical, curved organic subunits self-assemble through hydrogen bonds to form a closed spheroidal cavity capable of encapsulating guest molecules. The convergence of this geometric design principle across both organic hydrogen-bonded systems and inorganic MOPs underscores its universality for constructing closed-shell nanostructures from pre-organized, curved building blocks. As such, the overall assembly can be topologically represented as a truncated octahedron with the point symbol (46.68) (Fig. 2c). It is noteworthy that the square faces in the topological representation are realized as rectangles in the actual structure; this geometric distortion does not affect the topological classification. To the best of our knowledge, this structure provides an instance of a tro-type topology in Ti-MOP (Supplementary Table S3). The resulting structure measures approximately 18.7 × 18.7 × 20.2 Å3 and features a spherical inner cavity with a diameter of roughly 10.0 Å. Electrostatic potential (ESP) calculations reveal that FIR-151 is electrically neutral overall, yet features negatively charged open windows that favor interaction with cationic guest species (Supplementary Fig. 6). Moreover, these cages adopt a body-centered cubic close-packed arrangement in FIR-151 (Supplementary Fig. 7).

To deconvolute the pathway leading to this MOP, we designed a strategy to perturb the assembly equilibrium using kinetic trappers, aiming to stabilize and isolate metastable intermediates for structural characterization (Supplementary Fig. 8).

Intervention I: Trapping a macrocyclic precursor with Ni2+. Introducing NiCl2·6H2O into the reaction mixture leads to the isolation of yellow block-shaped crystals of FIR-152 (Supplementary Fig. 9). SCXRD analysis reveals that it crystallizes in the monoclinic space group P21/n. Its structure is represented by the formula [Ni(CH3CN)6]2+·[Ti12Ni6O12Cl8(C4O4)6(CH3COO)18(CH3CN)6]2– (Supplementary Table S4), and is built around Ti12 macrocycle—corresponding to intermediate 1 (Fig. 2d). Two Ni3Cl4 clusters are positioned above and below the Ti12 macrocycle and are connected to it via three C4O42– ligands, forming a heterometallic MOP, FIR-152. The incorporation of the two Ni3Cl4 clusters (Fig. 2e) results in a chair-like configuration of the Ti12 macrocycle ring. We propose that Ni2+ acts as a kinetic trap for this oligomeric species by modulating the local coordination thermodynamics, thereby stabilizing a critical nucleation intermediate along the assembly pathway.

Intervention II: Directing evolution with quaternary ammonium ions. When tetrabutylammonium bromide (TBAB) was introduced under otherwise identical reaction conditions, a structural rearrangement occurred, leading to the formation of orange hexagonal dipyramid crystals assigned as the isomeric phase FIR-153 (space group Cmcm, Supplementary Fig. 10 and Table S4). This compound features a Ti12 (6-4-6) modular structure (intermediate 2, Fig. 2f) in which the original 12-membered ring adopts a boat conformation. A central C4O42– ligand partitions the ring via coordination to four Ti4+ ions, while two Ni3Cl4 clusters reposition to apical sites, linked by additional C4O42– ligands to form an isomeric MOP. Here, Ni2+ again serves to kinetically trap this reconfigured module. This transformation demonstrates how non-coordinating additives can profoundly influence the assembly landscape, steering the reorganization of a metastable intermediate into an on-pathway structural motif preceding the final MOP. The molecular structures of both intermediates, FIR-152 and FIR-153, were corroborated in solution by 1H NMR and small-angle X-ray scattering (SAXS) techniques (Supplementary Figs. 11–14). To elucidate the assembly pathway of FIR-151, we employed a combination of solution-phase, structural, and reactivity analyses. Real-time SAXS monitoring revealed a continuous evolution of the maximum particle dimension (Dmax), as derived from the pair distance distribution function, P(r), from an initial ~6.7 Å through discrete states (~9.5 Å, ~17.0 Å) to the final ~19.1 Å (Supplementary Fig. 15). This progression is consistent with a multi-step assembly process rather than a single-step reaction. To gain atomic-level insight into potential intermediates along this pathway, we introduced Ni2+ ions as a kinetic trap, successfully isolating two key structural modules: the Ti12 macrocycle (FIR-152) and the Ti12 (6-4-6) module (FIR-153). Their precise structural complementarity provides a plausible geometric model for the progression towards the final cage. Critical reactivity tests further support this interpretation: while FIR-151 forms efficiently in the absence of Ni2+, the Ni2+-stabilized FIR-153 cannot be converted to the final product under identical conditions. This indicates that Ni2+ acts to trap and stabilize transient, on-pathway species, which would otherwise be unobservable. While each line of evidence—solution dynamics, structural correlation, and chemical reactivity—has inherent limitations, their convergence mutually supports a coherent, hierarchical assembly pathway from simple precursors to the complex Ti24 architecture.

Comparative analysis of the intermediates and the final product with precise crystal structures allows us to propose a hierarchical assembly pathway: Ti12 macrocycle (intervention 1) → Reorganization → Ti12 (6-4-6) module (intervention 2) → Dimerization → Ti24 MOP (Fig. 3a). This mechanistic proposal is best regarded as a robust working hypothesis, substantiated by the convergence of multiple experimental observations within the current technical constraints. We acknowledge that definitive proof would require direct molecular-level tracking of the transient intermediates in solution. Nevertheless, this hypothesis provides a valuable structural and conceptual framework for understanding the complex self-assembly of high-nuclearity MOPs and paves the way for future investigations using advanced in situ techniques.

Fig. 3: Energy landscape and the role of coordination gradient.
Fig. 3: Energy landscape and the role of coordination gradient.
Full size image

a A simplified energy landscape of the assembly of the Ti24 MOP. The reaction pathway involves i) kinetically trapped intermediates (intermediate 1 and intermediate 2) and ii) a high kinetic barrier corresponding to the dimerization step. The Ni2+ ion acts to stabilize the reaction energy, facilitating intermediate capture, while the dynamic nature of the Ti4+ ion promotes structural reorganization along the pathway. b Coordination lability gradient directs assembly pathway. The Ti4+ ion, a hard acid, exhibits high lability with rapidly exchanging coordination bonds (dotted line), driving structural reorganization. In contrast, the borderline acid Ni2+ ion displays intermediate lability, forming more stable coordination bonds that act as a kinetic trap and stabilize transient intermediates.

This work supplies an effective strategy that extends beyond a single case study: the strategic use of coordination lability gradients (Fig. 3b). The assembly is controlled by the differential kinetics between the highly labile Ti4+ ion, which drives dynamic reorganization, and the moderately labile Ni2+ ion, which acts as a kinetic trap to stabilize key intermediates at defined points along the pathway. This strategy of using a second metal ion to modulate the energy landscape of a primary assembly process offers an alternative approach to achieve structural complexity, and its broad utility awaits validation in other metal-organic systems.

The as-synthesized FIR-151 were further characterized using various analytical techniques to confirm their phase purity, thermal stabilities, optical bandgap, and photoactivity (Supplementary Figs. 16–22). FIR-151 exhibits high stability towards common organic solvents and some acids and bases (Supplementary Fig. 23), enabling investigation of its gas adsorption and separation properties. To evaluate its porosity, the solvent-exchanged sample (CH3CN) was activated under vacuum at 30 °C for 12 h (Supplementary Fig. 24). N2-sorption isotherms measured at 77 K revealed a type I isotherm, confirming the microporous nature and permanent porosity of FIR-151 (Fig. 4a). The Brunauer–Emmett–Teller (BET) and Langmuir surface areas were calculated to be 703 and 859 m2·g–1 (Supplementary Fig. 25), respectively—significantly higher than those of most reported MOPs2,46. The total pore volume, estimated from the N2 uptake at a relative pressure of P/P0 = 0.998, is 0.322 cm3·g–1 (Supplementary Fig. 26). Density functional theory (DFT) pore size distribution analysis further indicated a predominant pore width of 7.5 Å (Fig. 4b).

Fig. 4: Porosity and gas separation selectivity of FIR-151.
Fig. 4: Porosity and gas separation selectivity of FIR-151.
Full size image

a N2-sorption isotherm at 77 K for FIR-151. b Pore-size distribution curve of FIR-151 derived from density functional theory (DFT) analysis. c The sorption isotherms of FIR-151 for CO2, C2H2, C2H4, and C2H6 at 273 K. d The sorption isotherms of FIR-151 for CO2, C2H2, C2H4, and C2H6 at 298 K. e The IAST selectivities of FIR-151 for C2H2/C2H6 (50/50, v/v) mixtures at 1 bar. f The IAST selectivities of FIR-151 for CO2/C2H6 (50/50, v/v) mixtures at 1 bar. Source data are provided as a Source Data file.

Leveraging its high surface area, we further assessed FIR-151’s gas separation potential via CO2, C2H2, C2H4, and C2H6 sorption at 273 and 298 K (Fig. 4c, d, and Supplementary Figs. 27–31). FIR-151 exhibits a high CO2 uptake capacity of 79 cm3·g–1 at 273 K and 51 cm3·g–1 at 298 K, with an isosteric heat of adsorption (Qst) of 40 kJ·mol–1, indicating strong host-guest interactions within the pores. Furthermore, FIR-151 shows distinct gas uptake profiles for C2 hydrocarbons: under identical conditions, sorption capacities for C2H2, C2H4, and C2H6 were measured as 73, 53, and 35 cm3 · g–1 at 273 K, and 60, 38, and 25 cm3·g–1 at 298 K, respectively. The corresponding Qst values are 32, 28, and 27 kJ·mol–1, reflecting differential binding affinities. The notable contrast in both uptake capacity and adsorption enthalpy among these gases underscores the potential of FIR-151 for highly selective gas separation. To quantitatively predict separation selectivity, ideal adsorbed solution theory (IAST) was employed based on the above adsorption results (Supplementary Figs. 32–37). As a result, the predicted selectivity for an equimolar C2H2/C2H6 (v/v = 50:50) mixture reaches 42 at 273 K and 53 at 298 K (Fig. 4e). In addition, FIR-151 also shows favorable selectivities for CO2/C2H6 (v/v = 50:50; 15 at 273 K and 9 at 298 K (Fig. 4f), outperforming most reported MOPs46,47,48. Furthermore, the structural integrity and permanent porosity of FIR-151 were confirmed by powder X-ray diffraction (PXRD) measurements after gas sorption and the reproducibility of two consecutive BET measurements (Supplementary Fig. 38). These results demonstrate how mechanistic understanding and controlled synthesis of MOPs can yield functional materials49.

Building on the structural insights of FIR-151, we explored its post-assembly modification potential to tailor material properties (Fig. 5a). Specifically, the replacement of acetate ligands in FIR-151 with methacrylate and benzoate to form FIR-154 and FIR-155, respectively. The successful ligand exchange was verified by 1H NMR spectroscopy (Supplementary Figs. 39, 40), while the retention of the intact MOP architecture in solution was confirmed by SAXS (Supplementary Figs. 41 and 42). Furthermore, PXRD spectra confirmed the phase purity of the modified MOPs in the solid state (Supplementary Figs. 43 and 44). SCXRD confirmed that both compounds retain the structural integrity of the parental Ti24 MOP (Supplementary Table S5). Furthermore, all three MOPs adopt a body-centered cubic close-packed arrangement (Fig. 5b–d). The introduction of benzoate groups in FIR-155 enhances the material’s hydrophobicity, as confirmed by contact angle measurements (Supplementary Fig. 45), and facilitates π-π stacking (3.68 Å) between MOPs in its body-centered cubic lattice. Concurrently, the incorporation of methacrylate groups in FIR-154 installs structurally confirmed polymerizable handles, establishing a rational molecular design for exploring chemical functionality tuning and serving as a building block for extended polymeric networks in future studies.

Fig. 5: Post-assembly modification of Ti24 MOPs.
Fig. 5: Post-assembly modification of Ti24 MOPs.
Full size image

a Photographs and schematic of surface ligand exchange on Ti24 MOPs, where acetate ligands (CH3COO–, FIR-151) are replaced by methacrylic acid (CH2=C(CH3)COO–, FIR-154) and benzoic acid (C6H5COO–, FIR-155), scale bar = 50 μm. Crystal packing of modified Ti24 MOPs viewed along the c-axis: b FIR-154, c FIR-151, and d FIR-155. Atom colors: Ti sky blue, C gray, O red. The green dashed lines indicate hydrogen bonds or π-π interactions between adjacent MOPs.

Discussion

In summary, this study significantly advances the field of titanium supramolecular chemistry through the synthesis and structural characterization of FIR-151, a Ti24 MOP that exhibits a rare tro topology and represents the highest nuclearity Ti-MOP reported to date. The isolation and structural resolution of two key intermediates, a Ti12 macrocycle in FIR-152 and a Ti12 (6-4-6) module in FIR-153, provide critical insight into a hierarchical assembly pathway governed by differential coordination kinetics. FIR-151 demonstrates permanent porosity, selective gas sorption behavior, and post-assembly modification, underscoring its potential as a versatile platform for applied supramolecular science.

By leveraging the inherent reactivity of Ti4+ ions through rational pathway engineering, we transform a long-standing synthetic challenge into a strategic advantage for constructing architecturally complex supramolecular systems. This approach establishes an effective strategy for the rational construction of high-nuclearity MOPs, with potential applications in supramolecular synthesis, host-guest chemistry, and functional materials design.

Methods

Synthesis of FIR-151

A mixture of Ti(OiPr)4 (0.325 mmol, 100 μL), H2C4O4 (0.1 mmol, 0.011 g), CH3CN (4 mL), and CH3COOH (1 mL) in a 25 mL Teflon-lined stainless-steel autoclave. Then, the mixture was heated at 120 °C for 3 days. After cooling to room temperature, the yellow rhombic dodecahedral crystals were washed with a large amount of CH3CN, filtered, and stored in CH3CN (Yield: ∼13.1 mg, ca. 22% based on Ti). One of these crystals was used for X-ray crystallography.

Synthesis of FIR-152

A mixture of Ti(OiPr)4 (0.325 mmol, 100 μL), NiCl2·6H2O (0.2 mmol, 0.048 g), H2C4O4 (0.1 mmol, 0.011 g), CH3CN (4 mL), and CH3COOH (1 mL) in a 20 mL vial. Then, the mixture was heated at 80 °C for 7 days. After cooling to room temperature, the yellow block-shaped crystals were washed with CH3CN and stored in CH3CN (Yield: ∼20.0 mg, ca. 33% based on H2C4O4). One of these crystals was used for X-ray crystallography.

Synthesis of FIR-153

A mixture of Ti(OiPr)4 (0.325 mmol, 100 μL), NiCl2·6H2O (0.2 mmol, 0.048 g), H2C4O4 (0.1 mmol, 0.011 g), tetrabutylammonium bromide (TBAB, 0.2 mmol, 0.062 g), CH3CN (4 mL), and CH3COOH (1 mL) in a 20 mL vial. Then, the mixture was heated at 80 °C for 7 days. After cooling to room temperature, the orange hexagonal dipyramid crystals were washed with CH3CN and stored in CH3CN (Yield: ∼21.2 mg, ca. 35% based on H2C4O4). One of these crystals was used for X-ray crystallography.

Synthesis of FIR-154

FIR-151 crystals (0.005 g, 1.154 × 10-6 mol) were dissolved in a DMSO/CH3CN mixture (v/v = 1/9, 1 mL) under ultrasonication. Methacrylic acid (0.30 mmol, 25 μL) was then added, and the solution was heated at 80 °C for 7 days. After cooling to room temperature, orange crystals of FIR-154 were collected by filtration, washed with CH3CN, and stored in CH3CN (Yield: ∼5.4 mg, ca. 89% based on FIR-151). One of these crystals was used for X-ray crystallography.

Synthesis of FIR-155

FIR-151 crystals (0.005 g, 1.154 × 10-6 mol) were dissolved in a DMSO/CH3CN mixture (v/v = 1/9, 1 mL) under ultrasonication. Benzoic acid (0.30 mmol, 0.036 g) was then added, and the solution was heated at 80 °C for 7 days. After cooling to room temperature, yellow single crystals of FIR-155 were collected by filtration, washed with CH3CN, and stored in CH3CN (Yield: ∼6.5 mg, ca. 86% based on FIR-151). One of these crystals was used for X-ray crystallography.

Solvent exchange and activation of FIR-151

Freshly prepared FIR-151 was soaked in CH3CN at room temperature for 7 days to exchange the guest CH3COOH. Subsequently, the samples were activated under dynamic vacuum at 30 °C for 12 h. This temperature ensured the complete removal of the solvents, meanwhile excluding the possibility of FIR-151 decomposition.