Backbone‐Bridging Promotes Diversity in Heteroleptic Cages

Abstract The combination of shape‐complementary bis‐monodentate ligands LA and LB with PdII cations yields heteroleptic cages cis‐[Pd2 LA 2 LB 2] by self‐sorting. Herein, we report how such assemblies can be diversified by introduction of covalent backbone bridges between two LA units. Together with solvent and guest effects, the flexibility of these linkers can modulate nuclearity, topology, and number of cavities in a family of four structurally diverse assemblies. Ligand LA1, with flexible linker, reacts in CH3CN with its LB counterpart to a tetranuclear dimer D1. In DMSO, however, a trinuclear pseudo‐tetrahedron T1 is formed. The product of LA2, with rigid linker, looks similar to D1, but with a rotated ligand arrangement. In presence of an anionic guest, this dimer D2 transforms and a hexanuclear prismatic barrel P2 crystallizes. We demonstrate how controlling a ligand's coordination mode can trigger structural differentiation and increase complexity in metallo‐supramolecular assembly.


Materials and measurements
Unless otherwise stated, all chemicals were obtained from commercial sources and used as received.

Self-assembly of homoleptic ring/tetrahedron mixture [Pd3(L B )6] 6+ (R)/[Pd4(L B )8] 8+ (T) in DMSO-d6
To a solution of L B (7 mM, 2.1 μmol) in 447 μL DMSO-d6 was added a stock solution of [Pd(CH3CN)4](BF4)2 (53 μL, 20 mM/DMSO-d6, 1.05 μmol). The mixture was kept at r.t. for 3 h to give a mixture of homoleptic ring/tetrahedron (R/T).             Figure S37. 13 C NMR spectrum (151 MHz, 298K, DMF-d7) of heteroleptic cage dimer D2. Figure S38. 1 H NMR spectra (500 MHz, 298K, DMSO-d6) of the mixture of homoleptic bridged cage C1 and a 1:1 mixture of homoleptic ring R and tetrahedron T at room temperature over the course of 12h. As can be seen, homoleptic species co-exist and no conversion to heteroleptic pseudo-tetrahedron T1 is achieved at this temperature. This experiment shows that the heteroleptic pseudo-tetrahedron T1 is stable in solution up to 378 K. Moreover, the split proton signals do not coalesce even at elevated temperatures, indicating that the signals splitting is caused by an inherently low symmetry of the topology (not by conformational locking effects), further supporting the assignment to species T1 instead of R1.
As the host-guest interactions between cage dimer D1 and G 1 were observed to follow slow exchange kinetics, the concentrations of G 1 @D2, 2G 1 @D2 and D2 could all be estimated from the 1 H NMR spectroscopic results. The integrals of protons a and b of ligand L B were used to approximate K1 and K2 by using the following equations: where H and G represent host cage dimer D2 and guest G 1 , respectively. Concentrations obtained from three distinct 1 H NMR spectra (0.6, 0.8 -1.0 eq) are tabulated in Table S1. Further, from this data, the cooperativity parameter α = 4K2/K1 was calculated. [7]

Ion Mobility Mass Spectrometry
Ion mobility measurements were performed on a Bruker timsTOF instrument combining a trapped ion mobility (TIMS) with a time-of-flight (TOF) mass spectrometer in one instrument. In contrast to the conventional drift tube method to determine mobility data, where ions are carried by an electric field through a stationary drift gas, the TIMS method is based on an electric field ramp to hold ions in place against a carrier gas pushing them in the direction of the analyzer. Consequently, larger sized ions that experience more carrier gas impacts leave the TIMS units first and smaller ions elute later. This method offers a much higher mobility resolution despite a smaller device size.   [8] (tCCS) based on the CREST [9] (GFN2-xTB) generated models with corresponding number of encapsulated BF4 − counter anions.

Species eCCS [Å²] tCCS (T1) [Å²] ∆% (T1) tCCS (R1) [Å²] ∆% (R1)
[ Experimental and theoretical data show in accordance that the CCS decreases with increasing number of encapsulated BF4 − counter anions, leading to a stepwise decrease of overall charge. This common phenomenon can be explained by weaker ion-induced dipole and ion-quadrupole interactions with the carrier gas molecules (N2), [10] and this trend is reproducible by the theoretical calculations. The same observation was also made for the heteroleptic dimer (Table S3).    [8] (tCCS) based models, which were optimized using B97-3c (ORCA, ver. 4.2.1) [11] , with corresponding encapsulated BF4 − counter anions. (peak with four counter anions chosen to allow direct comparison), which in turn has a similar CCS value compared to the species with only one guest bound. Figure S45. Model of heteroleptic cage C optimized with B3LYP/def2-SVP (ORCA 4.2.1) [11] in different views, a) side view, b) top view (a methyl group was used for ligand L A instead of its hexyl group to reduce calculation time. Two BF4 − counter anions inside the cavity are omitted for clarity). Figure S46. Model of bridged homoleptic cage C1, optimized with B3LYP/def2-SVP (ORCA 4.2.1) [11] in different views, a) side view, b) top view (two BF4 − counter anions inside the cavity are omitted for clarity).  [11] . A subsequent single point energy calculation shows that T1 is energetically more favourable than R1 with a difference of 24 kJ/mol. Colours indicate different chemical environments for substructures of ligands L A1 and L B .

S32
6. X-ray Crystallography Yellow block-shaped crystals of [Pd4(L A1 )2(L B )4](BF4)8 (D1) were grown by slow vapor diffusion of Et2O into a solution of the assembly product of L A1 and L B in CD3CN. Data was collected in-house on a Bruker D8 venture diffractometer equipped with an INCOATEC microfocus sealed tube (Iμs 3.0) using CuKα radiation at 100 K. The data was integrated with APEX3 and the structure was solved by intrinsic phasing/direct methods using SHELXT [12] and refined with SHELXL [13] for full-matrix least-squares routines on F 2 and ShelXle [14] as a graphical user interface and the DSR [15] program plugin was employed for modeling.

Specific refinement details of D1.
Stereochemical restraints for the ligands (L A1 and L B ) were generated by the GRADE program using the GRADE Web Server (http://grade.globalphasing.org) and applied in the refinement. A GRADE dictionary for SHELXL contains target values and standard deviations for 1,2-distances (DFIX) and 1,3-distances (DANG), as well as restraints for planar groups (FLAT). All displacements for non-hydrogen atoms were refined anisotropically. The refinement of ADP's for carbon, nitrogen and oxygen atoms was enabled by a combination of similarity restraints (SIMU) and rigid bond restraints (RIGU). The contribution of the electron density from disordered counterions and solvent molecules, which could not be modeled with discrete atomic positions were handled using the SQUEEZE routine in PLATON. The solvent mask file (.fab) computed by PLATON was included in the SHELXL refinement via the ABIN instruction leaving the measured intensities untouched.    .0) using CuKα radiation at 100 K. The data was integrated with APEX3 and the structure was solved by intrinsic phasing/direct methods using SHELXT [12] and refined with SHELXL [13] for full-matrix least-squares routines on F 2 and ShelXle [14] as a graphical user interface and the DSR [15] program plugin was employed for modeling.

Specific refinement details of D2.
Stereochemical restraints for the ligands (L A2 and L B ) were generated by the GRADE program using the GRADE Web Server (http://grade.globalphasing.org) and applied in the refinement. A GRADE dictionary for SHELXL contains target values and standard deviations for 1,2-distances (DFIX) and 1,3-distances (DANG), as well as restraints for planar groups (FLAT). All displacements for non-hydrogen atoms were refined anisotropically. The refinement of ADP's for carbon, nitrogen and oxygen atoms was enabled by a combination of similarity restraints (SIMU) and rigid bond restraints (RIGU). The contribution of the electron density from disordered counterions and solvent molecules, which could not be modeled with discrete atomic positions were handled using the SQUEEZE routine in PLATON. The solvent mask file (.fab) computed by PLATON was included in the SHELXL refinement via the ABIN instruction leaving the measured intensities untouched.    Symmetry code: $2=+x, 1-y, +z $3=1-x, +y, -z $4=1-x, 1-y, -z S37 6.3. Crystal structure of P2 Colorless needle-shaped crystals of [2G 1 +Pd6(L A2 )3(L B )6](BF4)8 (P2) were grown by slow vapor diffusion of Et2O into a solution of 2G 1 @D2 in DMF. Data was collected in-house on a Bruker D8 venture diffractometer equipped with an INCOATEC microfocus sealed tube (Iμs 3.0) using CuKα radiation at 100 K. The data was integrated with APEX3 and the structure was solved by intrinsic phasing/direct methods using SHELXT [12] and refined with SHELXL [13] for full-matrix least-squares routines on F 2 and ShelXle [14] as a graphical user interface and the DSR [15] program plugin was employed for modeling 6.3.1. Specific refinement details of P2.
Stereochemical restraints for the ligands (L A2 and L B ) were generated by the GRADE program using the GRADE Web Server (http://grade.globalphasing.org) and applied in the refinement. A GRADE dictionary for SHELXL contains target values and standard deviations for 1,2-distances (DFIX) and 1,3-distances (DANG), as well as restraints for planar groups (FLAT). All displacements for non-hydrogen atoms were refined anisotropically. The refinement of ADP's for carbon, nitrogen and oxygen atoms was enabled by a combination of similarity restraints (SIMU) and rigid bond restraints (RIGU). The contribution of the electron density from disordered counterions and solvent molecules, which could not be modeled with discrete atomic positions were handled using the SQUEEZE routine in PLATON. The solvent mask file (.fab) computed by PLATON was included in the SHELXL refinement via the ABIN instruction leaving the measured intensities untouched. Figure S56. Atomic numbering scheme of residue N7S (guest molecule G 1 ). Figure S57. Crystal structure of cage trimer P2 showing one G 1 positioned outside the cage boundaries and its hydrogenbonding environment (O⋯H separations are given in Ångström). Hydrogens, BF4anions and other solvent molecules are omitted for clarity.