A cage-on-MOF strategy to coordinatively functionalize mesoporous MOFs for manipulating selectivity in adsorption and catalysis

Metal–organic frameworks (MOFs) with large cavities offer high surface areas and open channels favorable for transport and hosting of guests, but unmodified MOFs often lack selectivity for substrates of similar size or properties, leading to side products and compromised separations. Conventional strategies (bottom-up synthesis or post-synthetic modification) create substrate-specific voids but commonly rely on nonporous surface capping agents that can reduce internal porosity or on noncovalent cage encapsulation that may block pores and be unstable under harsh conditions. The research question addressed is whether porous molecular cages can be coordinatively attached to MOF external surfaces to modulate surface charge, recognition, and catalytic behavior while preserving intrinsic porosity. The study presents a "Cage-on-MOF" approach using oppositely charged PCCs with secondary coordination sites to functionalize mesoporous MOFs (PCN-222 and MIL-101) and evaluates effects on adsorption selectivity and catalytic product selectivity.

Prior work has incorporated various surface capping agents (polymers, phospholipids, phenylsilanes, biomacromolecules) onto MOFs to improve dispersibility, stability, and separations. Encapsulation of macrocycles and cages (e.g., cucurbit[6]uril, metal–organic polyhedra) within MOF cavities enhanced CO2 adsorption and stability but often employed noncovalent interactions and placed nonporous or small cages inside pores, which can compromise porosity and raise stability concerns under extreme conditions. Previous PCCs from the authors’ group (anionic PCC-2b and cationic PCC-3) showed strong guest inclusion and catalytic properties but lacked secondary coordination sites to anchor onto MOF surfaces. Literature also shows sulfate- and amine-bearing ligands can coordinate to exposed metal sites in MOFs, suggesting a route to covalently (coordinatively) bind porous cages at external surfaces to tailor function without blocking internal pores.

• Design and synthesis of porous coordination cages: two PCCs with secondary coordination groups—PCC-4 (Co4(μ4-OH) clusters bearing sulfate groups, overall anionic) and PCC-5 (Pd(II) nodes bearing amino groups, overall cationic)—were synthesized and structurally characterized by single-crystal X-ray diffraction, 1H NMR, and ESI-MS. Previously reported PCC-2b (anionic) and PCC-3 (cationic) lacking secondary coordination groups served as controls.
• MOF selection and preparation: Mesoporous MOFs PCN-222 (Zr6 clusters with porphyrin linkers) and MIL-101 (Cr-based) were synthesized as nanoparticles (PCN-222 ~200 nm × 50 nm) and used as hosts.
• Cage-on-MOF assembly: As-synthesized MOF nanoparticles were dispersed in PCC solutions and stirred at 50 °C for 12 h (solvents: DMF or MeCN depending on PCC). Products (PCN-222@PCC-4, PCN-222@PCC-5; MIL-101@PCC-4, MIL-101@PCC-5) were isolated by centrifugation and washing to remove unbound cages. Control composites with PCC-2b or PCC-3 were prepared analogously.
• Characterization: TEM/SEM and HAADF-STEM with EDX elemental mapping and line scans assessed morphology and spatial distribution of cage metals (Co, Pd). PXRD verified crystallinity. Zeta potential measured surface charge changes. FT-IR identified cage functional groups on MOF surfaces. XPS probed binding energy shifts (Zr 3d for PCN-222; Cr 2p for MIL-101). N2 adsorption at 77 K (post-activation) quantified porosity/BET area; ICP-OES quantified cage loading and surface density. Molecular dynamics simulations assessed cage localization (outside vs inside pores). Ball-milling and probe sonication (MIL-101 composites) physically removed exterior layers to test surface localization via evolving Co/Cr and Pd/Cr ratios.
• Adsorption studies: Selective dye adsorption evaluated in water using cationic methylene blue (MB) and anionic Eosin Y (EY) for PCN-222 systems (single- and mixed-dye solutions; UV–vis monitoring; kinetics 0–120 min). For MIL-101 systems, Rhodamine B and methyl orange were used. FT-IR confirmed dye adsorption; reversibility tested via elution.
• Catalysis: Sequential reaction—hydrolysis of benzaldehyde dimethyl acetal followed by Knoevenagel condensation with malononitrile—was performed using PCN-222-based catalysts (50 °C, 5 h) and MIL-101-based catalysts (room temp, 2 h). Yields and selectivity quantified by GC. Recyclability assessed over 10 runs (morphology, PXRD, ICP-OES for cage retention).
• Mechanistic tests: Cavity-blocking inhibitors—tetraphenylphosphonium chloride for PCC-4 and 1-adamantane carboxylic acid for PCC-5—were used either pre-loaded into cages or added in situ to probe cavity involvement in selectivity. Time-dependent NMR and GC tracked intermediates. DFT calculations evaluated adsorption energies and binding modes of protonated benzaldehyde intermediates with PCC-4 and PCC-5.
• Reversible modification: Acid treatment (1 M HCl, ambient, 2 h) removed surface-coordinated cages from MOFs, enabling sequential re-installation of the opposite-charge PCC to reversibly tune catalytic selectivity; integrity verified by microscopy, mapping, and diffraction.

• Successful surface coordination of porous cages: STEM-EDX mapping showed uniform Co (PCC-4) or Pd (PCC-5) signals across MOF particles; line-scan volcano-shaped Zr/PCC ratios indicated surface localization. MD simulations showed negligible penetration of PCC-4/PCC-5 into PCN-222 pores due to sterics/electrostatics.
• Preserved porosity: PCN-222@PCCs retained type VI N2 isotherms with only ~5.6% capacity decrease at high P and similar BET areas to pristine PCN-222. In contrast, non-coordinating controls (PCC-2b, PCC-3) reduced N2 uptake by ~36%, evidencing pore blocking upon encapsulation.
• Quantified loadings and surface density: For PCN-222 composites, ICP-OES gave PCC loadings of 5.40 wt% (PCC-4) and 1.98 wt% (PCC-5), corresponding to surface-loaded PCC/Zr6 ratios of 10.50% and 8.83%. Controls showed much lower loadings (~1.06 wt% PCC-2b; 0.81 wt% PCC-3). For MIL-101, loadings were 23.5 wt% (PCC-4) and 8.2 wt% (PCC-5); theory suggested coordination to 13.3% and 9.1% of surface Cr3O clusters.
• Tuned surface charge: PCN-222@PCC-4 reversed surface charge to −20 mV; PCN-222@PCC-5 increased to +26 mV. For MIL-101, zeta potential shifted from +40 mV (pristine) to −22 mV (PCC-4) and to +51 mV (PCC-5).
• Enhanced stability: MOF@PCCs exhibited improved photo/chemical/thermal stability vs pristine MOFs (e.g., sustained PXRD under harsh conditions; photostability in benzylamine/methanol).
• Selective dye adsorption: For MB (cationic), maximum capacities (0–50 mg/L) were 175 mg/g (PCN-222@PCC-4), 148 mg/g (PCN-222), and 125 mg/g (PCN-222@PCC-5). In mixed MB/EY solutions, early-time selectivity (2 min) was 63:37 MB:EY for PCN-222@PCC-4 and 35:65 for PCN-222@PCC-5, while PCN-222 showed no selectivity; total dye uptake remained comparable to pristine MOF.
• Product-selective catalysis (PCN-222 systems): Pristine PCN-222 produced benzaldehyde (3) and benzylidenemalononitrile (4) in roughly equal amounts (no selectivity). PCN-222@PCC-4 favored product 4 (75.4% yield and selectivity), whereas PCN-222@PCC-5 favored intermediate 3 (89.7% yield and selectivity). The composites were recyclable with ~15.1% activity decline after 10 cycles, consistent with ~15.2% cage loss.
• Mechanistic support: Cavity inhibitors reduced selectivity by 27.8% (PCC-4) and 19.3% (PCC-5), implicating cage cavities. DFT: PCC-4 strongly adsorbed protonated benzaldehyde (Eads −3.7, −4.3, −3.3 eV), whereas PCC-5 showed weak/repulsive adsorption (Eads +3.0, +2.9 eV), rationalizing opposite selectivity.
• Reversibility: Acid treatment removed 53.8–67.7% of surface PCCs from PCN-222@PCC, enabling reinstallation of alternate PCC and reversible switching of product selectivity correlated with zeta potential.
• Generality (MIL-101): MIL-101@PCCs preserved crystallinity/porosity and showed surface localization validated by ball-milling/sonication (Co/Cr ratio drop 0.11→0.026; Pd/Cr 0.12→0.021). In sequential condensation, MIL-101@PCC-4 yielded phenylmethylene malononitrile with 64.9% yield and 100% selectivity; MIL-101@PCC-5 produced benzaldehyde diacetal with 62.1% yield and 88% selectivity. Selectivity was reversibly tunable by cage exchange via mild acid treatment.

The study demonstrates that coordinatively attaching porous cages to the external surfaces of mesoporous MOFs can tailor interfacial microenvironments—surface charge, binding interactions, and substrate recognition—without compromising intrinsic porosity. This addresses the central challenge of achieving selectivity while maintaining accessible voids. Electrostatic complementarity and selective adsorption of reaction intermediates within cage cavities drive reversed product selectivity in sequential reactions. Inhibitor-blocking experiments and DFT adsorption energy analyses support a mechanism where anionic PCC-4 strongly stabilizes protonated benzaldehyde, favoring progression to the Knoevenagel product, whereas cationic PCC-5 disfavors this intermediate, arresting the sequence at benzaldehyde. The strategy is generalizable to different MOFs (PCN-222, MIL-101), enhances stability, and is operationally reversible, enabling on-demand tuning of heterogeneous catalytic outcomes by swapping surface-bound cages.

A Cage-on-MOF strategy was established to coordinatively graft oppositely charged porous coordination cages onto mesoporous MOFs, preserving porosity while enabling tunable adsorption and catalytic selectivity. The resulting MOF@PCCs show: (i) controlled surface charge and enhanced stability; (ii) selective dye recognition with maintained overall uptake; (iii) product-selective heterogeneous catalysis with excellent recyclability; and (iv) reversible switching of selectivity via removable surface cages. This constitutes the first example of using a porous molecular cage as a surface-capping agent to covalently (coordinatively) functionalize porous frameworks for switchable surface charge and controllable product selectivity. The approach is general across MOFs and suggests future research into expanding cage libraries, host–cage combinations, and reaction scopes for programmable heterogeneous catalysis and separations.