Large, heterometallic coordination cages based on ditopic metallo-ligands with 3-pyridyl donor groups

The two-step synthesis of large, heterometallic coordination cages based on ditopic 3-pyridyl ligands is described.


General
All chemicals were obtained from commercial sources and used without further purification. Pd(NO 3 ) 2 (H 2 O) n was supplied as 40% Pd by weight. Solvents were dried using a solvent purification system from Innovative Technologies, Inc.. Reactions were carried out under an atmosphere of dry N 2 using standard Schlenk techniques. Routine 1 H, 19 F and 13 C NMR spectra were obtained on a Bruker Avance III spectrometer ( 1 H: 400 MHz, 19 F: 376 MHz, 13 C: 101 MHz) equipped with a 5 mm BBFO-Plus z probe. Low and elevated temperature 19 F spectra were obtained on a Bruker DRX spectrometer ( 1 H: 400 MHz, 19 F: 376 MHz) equipped with a 5 mm BBO probe, and DOSY and NOESY NMR spectra were recorded on a Bruker Avance spectrometer ( 1 H: 400 MHz) equipped with a 5 mm BBIz probe. 1 H chemical shifts are reported in parts per million δ (ppm) referenced to an internal solvent. 19 F chemical shifts are reported in ppm and referenced to external hexafluorobenzene in CD 3 CN (−164.90 ppm). 13 C chemical shifts are reported in ppm and referenced to an internal solvent. All routine, and DOSY and NOESY NMR spectra were recorded at 298K. Electrospray-ionization MS data were acquired on a Q-TOF Ultima mass spectrometer (Waters) operated in the positive ionization mode and fitted with a standard Z-spray ion source equipped with the Lock-Spray interface. Data were processed using the MassLynx 4.1 software.

Ligand 1
Dimethylglyoxime (708 mg, 6.10 mmol), pyridin-3-ylboronic acid (500 mg, 4.07 mmol) and anhydrous FeCl 2 (257 mg, 2.03 mmol) were suspended in MeOH (40 mL) and heated under reflux for 3 h under N 2 to give a deep red solution. This reaction mixture was allowed to cool to RT before the volume of solvent was reduced by half under vacuum. An orange precipitate formed upon the addition of Et 2 O (50 mL). This solid was collected by centrifugation before being dissolved in CHCl 3 , transferred to a separatory funnel, washed with saturated aqueous NaHCO 3 (3 x 100 mL) and water (100 mL), dried (MgSO 4 ), and filtered before solvent was removed under vacuum to give 1 as a red solid (1. 12

Ligand 2
Nioxime (1.73 g, 12.2 mmol), pyridin-3-ylboronic acid (1.00 g, 8.14 mmol) and anhydrous FeCl 2 (514 mg, 4.06 mmol) were dissolved in MeOH (80 mL) and heated under reflux for 4 h under N 2 . The deep red reaction mixture was concentrated to dryness on a rotary evaporator before being redissolved in a minimum volume of CH 2 Cl 2 , transferred to a separatory funnel, and washed with saturated aqueous NaHCO 3 (3 x 100 mL) and water (100 mL), dried (MgSO 4 ), and filtered before solvent was removed under vacuum. The resulting red powder was recrystallized from MeOH to give 2 as a crystalline solid (2.10 g,

Cage 6
In a vial, 2 (50 mg, 0.076 mmol) was suspended in MeCN (5 mL) before (Pd(NO 3 ) 2 (H 2 O) n (10 mg, 0.038 mmol) was added as a solution in H 2 O (1 mL). The resulting suspension was heated at 70 ºC for 2 h before being allowed to cool to RT. The slightly turbid deep red reaction mixture was filtered before Et 2 O (10 mL) was added and an orange precipitate

Monitoring the Formation of 3-6
The formation of 3-6 was monitored by 1 H NMR as follows. An NMR tube was charged with 1 or 2 (7.62 μmol), and CD 3 CN (0.5 mL, for the preparation of 3 and 4) or was added to this suspension and the 1 H NMR spectrum was immediately recorded. The NMR tube was subsequently heated at 70 ºC in an oil bath, removed at intervals and cooled to RT before the 1 H NMR spectrum was again recorded.

Variable Temperature 19 F NMR Spectra of 3 and 4
Figure S25. 19 F NMR spectrum of 3 recorded at selected temperatures in CD 3 CN. The reference signal was used to reference to scale the intensities of the BF 4 − anion peaks. Figure S26. 19 F NMR spectrum of 4 recorded at selected temperatures in CD 3 CN. The reference signal was used to scale the intensities of the BF 4 − anion peaks.

General
Intensity data were collected at the Swiss Norwegian beamline BM01A at the ESRF in Grenoble (France) on the Pilatus@SNBL kappa goniometer from Huber Diffraktionstechnik GmbH equipped with a Pilatus2M pixel detector from Dectris Ltd, on an Oxford Diffraction KM-4 CCD diffractometer, a Bruker APEX II CCD system, or a marx system. All data collections were performed at low temperature (100−140 K) using a Cryostream 700 Series from Oxford Cryosystems Ltd. Data integration was carried out using Crystalys Pro, 1 BYPASS, 2 EVALCCD, 3 automar 4 or XDS 5,6 in combination with autoPROC. 7 Multi-scan empirical absorption corrections were applied to the data using CrysAlis Pro or SADABS. 8 Data reduction was carried out using XPREP. 8 All structures were solved by direct methods or charge flipping using SIR, 9 SHELXT 10 or SUPERFLIP 11 and refined with SHELXL 10 using full-matrix least-squares routines on F 2 and ShelXle 12 as graphical user interface.
A series of carefully adapted macromolecular refinement techniques enabled us to successfully build and complete a molecular model. These methods already proved successful in previous cases of huge and complicated supramolecular structures with high solvent content. 13−15 All atoms were grouped into residues to enable addressing all atoms of repeating structural fragments with a single command. This grouping is required for the application of geometric restraint dictionaries generated with the Grade Program, which is part of BUSTER 16

Ligands 1 and 2 Ligand 1: Data collection and refinement details
The asymmetric unit only contains half of the clathrochelate ligand 1. It was modelled with rotational disorder using a free variable which resulted in an occupancy ratio of 0.659 (3) : 0.341 (3). Due to minor dynamic rotational disorder of chlorine atoms, which is reflected in their large ADPs, the carbon atom of the dichloromethane solvent had a significantly lower equivalent U value compared to chlorine atoms. The slightly too low goodness of fit most likely reflects a (tiny) overestimation of the diffraction data's sigma values.

Ligand 2: Data collection and refinement details
Due to the measurement on a single axis diffractometer, a completeness of only 92.9% for the triclinic space group P-1 could be achieved. Nevertheless, the overall structural quality is very good.

Cage 3: Data collection and refinement details
The X-ray diffraction measurements were carried out on the Swiss Norwegian beamline

Cage 5: Data collection and refinement details
The X-ray diffraction measurements were carried out on the Swiss Norwegian beamline

Cage 6: Data collection and refinement details
The X-ray diffraction measurements were carried out in-house and few reflections greater

Cage 7: Data collection and refinement details
The X-ray diffraction measurements were carried out in-house and few reflections greater than 1.4 Å resolution were observed. The structure was solved using charge flipping as implemented in the Superflip software. 11 The available data did not allow for exhaustive modelling of finer structural details like (restrained) anisotropic displacement parameters of the 1392 independent non-hydrogen, location and placement of hydrogens at any of the 21 modelled water molecules or describing parts of disorder with several atomic positions.

S29
There are a few short contacts. Apart from observed electron density supporting the molecular model (see Figure S32) there is however a large flexibility in these regions. It is very likely that there are several (dynamically disordered) conformations of the affected moieties, which cannot be properly modelled at this (1.4 Å) resolution.
There is a significant amount of thermal motion in the extremities of the molecule and This structure is similar to a macromolecular type of structure (e.g. proteins) in terms of weak scattering power and resulting data quality (e.g high R int and wR2) as well as unit cell volume and number of independent atoms, and the adapted macromolecular refinement techniques described in the general part were key to establish the connectivity of all modeled structural components with reasonably good precision.
Although CHECKCIF criteria are valid and necessary to assess the quality of small molecule structures, some of the criteria (e.g. requirement of atomic resolution) are unsuited for structures of the macromolecular domain and should be replaced with more sensible ones for supramolecular structures with these characteristics. In this context, it should also be mentioned that the current version of CHECKCIF installed on the IUCR web server (checkcif.iucr.org/) cannot cope with structures of this size. We have therefore only been able to run an updated local version of CHECKCIF routines on this structure.
This updated version of CHECKCIF has kindly been provided by Prof. Dr. Anthony L. Spek.
S30 Figure S31. Atomic representation of the asymmetric unit of 7 showing both cages, acetonitrile and water solvent molecules as well as tetrafluoroborate and tetraphenylborate counter ions.

Cavity Volume Calculations
Crystallographically determined octahedral cages of structures 3, 5, 6 and 7 were symmetry expanded (if required) and guests molecules as well as counter ions were removed.
Resulting inner cavities were calculated with VOIDOO, 24 using a primary grid and plot grid spacing of 0.2 Å and ten cycles of volume refinement. To prevent the probe from "escaping" the inner sphere through the large pores, the default water size probe radius of 1.4 Å was increased to 1.7 Å for all structures. This procedure result in smaller calculated volumes compared to using the default probe size, but allows for a systematic comparison of resulting cavity volumes.
For both independent cages of structure 7 the cavities are visualized as a 50% transparent blue iso surface inside the crystal. Molecular visualization was done using PyMol. 25 Figure S33. Cavity visualization for both octahedral cages in the cage 3 structure. Cavities are shown as a 50% transparent blue iso surface.

Figure S34
Cavity visualization for assembly 5. Cavity shown as a 50% transparent blue iso surface. Encapsulated solvent molecules as well as counterions have been omitted.

Figure S35
Cavity visualization for both assembly 6. Cavities are shown as a 50% transparent blue iso surface. Solvent molecules as well as counterions have been omitted. 3 + 1 equiv. NaBPh4 3 + 2 equiv. NaBPh4 3 + 3 equiv. NaBPh4 3 + 4 equiv. NaBPh4 Figure S37. Changes in the 1 H NMR spectrum of 3 (bottom, red) in CD 3 CN upon addition of NaBPh 4 (second from bottom, yellow). For clarity, the aromatic region has been scaled to a much higher intensity than the aliphatic region.

NaBPh4
3 + 1 equiv. NaBPh4 ppm Figure S38. Changes in the 1 H NMR spectrum of 3 (bottom, red) in 2:1 CD 3 CN:D 2 O upon addition of NaBPh 4 (second from bottom, yellow). For clarity, the aromatic region has been scaled to a much higher intensity than the aliphatic region.

NaBPh4
3 + 1 equiv. NaBPh4 ppm Figure S39. Changes in the 1 H NMR spectrum of 3 (bottom, red) in 1:1 CD 3 CN:CDCl 3 upon addition of NaBPh 4 (second from bottom, yellow). For clarity, the aromatic region has been scaled to a much higher intensity than the aliphatic region.

DOSY NMR Spectra
BPh4 -free BPh4 -bound Figure S40. Aromatic region of the DOSY NMR spectrum of 3 in CD 3 CN after addition of one equivalent of NaBPh 4 .

BPh4 -free
BPh4bound Figure S41. Aromatic region of the DOSY NMR spectrum of 3 in CD 3 CN after addition of two equivalents of NaBPh 4 .

Calculation of Association Constants
Association constants were calculated by integration of 1 H NMR signals of protons α, β and γ corresponding to bound and unbound BPh 4 − anion. The ratio of the integrals of the two species was used to determine their relative concentrations.