Anion Pairs Template a Trigonal Prism with Disilver Vertices

Here we describe the formation of a trigonal prismatic cage, utilizing 2-formyl-1,8-naphthyridine subcomponents to bind pairs of silver(I) ions in close proximity. This cage is the first example of a new class of subcomponent self-assembled polyhedral structures having bimetallic vertices, as opposed to the single metal centers that typically serve as structural elements within such cages. Our new cage self-assembles around a pair of anionic templates, which are shown by crystallographic and solution-phase data to bind within the central cavity of the structure. Many different anions serve as competent templates and guests. Elongated dianions, such as the strong oxidizing agent peroxysulfate, also serve to template and bind within the cavity of the prism. The principle of using subcomponents that have more than one spatially close, but nonchelating, binding site may thus allow access to other higher-order structures with multimetallic vertices.


S1 Experimental Procedures
Unless otherwise specified, all starting materials, solvents and reagents were used as supplied and without further purification. Flash column chromatography was performed using Silica Gel high purity grade (pore size 60 Å, 230-400 mesh particle size, Sigma-Aldrich). All anhydrous reactions were carried out in oven-dried glassware and under an inert atmosphere of nitrogen provided by a balloon. All reactions were stirred with magnetic followers. Brine refers to a saturated aqueous solution of sodium chloride. Centrifugation of samples was carried out using a Grant-Bio LMC-3000 low speed benchtop centrifuge.

Mass Spectrometry
Low-resolution electrospray ionisation (ESI) mass spectra were obtained on a Micromass Quattro LC mass spectrometer (cone voltage 20 eV; desolvation temp. 313 K; ionisation temp. 313 K) infused from a Harvard syringe pump at a rate of 4-10 µL min −1 . High-resolution ESI mass spectra were obtained by the EPSRC UK National Mass Spectrometry Facility at Swansea University using a Thermo Scientific LTQ Orbitrap XL hybrid ion trap-orbitrap mass spectrometer.

NMR Spectroscopy
NMR spectra were recorded on Bruker 400 Avance III HD Smart Probe, Bruker Avance 500 Cryo, Bruker 500 TCI-ATM Cryo and Bruker 700 TCI-ATM Cryo spectrometers. Chemical shifts (δ) for 1 H NMR spectra are reported in parts per million (ppm) and are reported relative to the solvent residual peak. DOSY experiments were carried out on a Bruker DRX-400 spectrometer. Coupling constants (J) are reported in hertz (Hz). The following abbreviations are used to describe signal multiplicity in 1 H, 13 C and 19 F NMR spectra: s: singlet, d: doublet, t: triplet, dd: doublet of doublets; dt: doublet of triplets; m: multiplet, br: broad. DOSY NMR experiments were performed on a 400 MHz Avance III HD Smart Probe NMR spectrometer. Maximum gradient strength was 6.57 G/cm A. The standard Bruker pulse program, ledbpgp2s, employing a stimulated echo and longitudinal eddy-current delay (LED) using bipolar gradient pulses for diffusion using 2 spoil gradients was utilised. Rectangular gradients were used with a total duration of 1.5 ms. Gradient recovery delays were 875-1400 μs. Individual rows of the S4 quasi-2D diffusion databases were phased and baseline corrected.

S8 X-ray Crystallography
Data were collected at Beamline I19 of Diamond Light Source employing silicon double crystal monochromated synchrotron radiation (0.6889 Å) with ω and ψ scans at 100(2) K. 1 Data integration and reduction were undertaken with Xia2. [2][3][4] Subsequent computations were carried out using the WinGX-32 graphical user interface. 5 Multi-scan empirical absorption corrections were applied to the data using the AIMLESS 6 tool in the CCP4 suite. 7 The structures were solved by direct methods using SHELXT 8 then refined and extended with SHELXL. 9 In general, non-hydrogen atoms with occupancies greater than 0.5 were refined anisotropically. Carbon-bound hydrogen atoms were included in idealised positions and refined using a riding model. Disorder was modelled using standard crystallographic methods including constraints, restraints and rigid bodies where necessary. Crystallographic data along with specific details pertaining to the refinement follow. Crystallographic data have been deposited with the CCDC (1913631)(1913632)(1913633)(1913634)(1913635).

Specific refinement details:
The crystals of [(PF6)1.65(OTf)0.35⸦Ag12L6]ꞏ9.35PF6ꞏ0.65OTfꞏ2.25MeCN were grown by diffusion of diethyl ether into an acetonitrile solution of [Ag12L6]ꞏ12OTf containing excess Bu4NPF6. The crystals employed immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data. Data were obtained to 0.84 Å resolution. The asymmetric unit was found to contain one complete Ag12L6 assembly and associated counterions and solvent molecules.
Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other and additional bond length restraints (DFIX) were applied to some sections of the organic ligands. Thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for silver to facilitate anisotropic stable refinement.

S58
One of the encapsulated anions was modelled as a full occupancy PF6 − . The other one was modelled as a disordered mixture of PF6 − and OTf − with occupancies of 0.65 and 0.35 respectively; these occupancies were initially refined and then fixed at the obtained values. The other anions within the structure also show evidence of substantial disorder. A further anion lattice site was modelled as a disordered mixture of PF6 − and OTf − and three PF6 − anions were modelled as disordered over two locations. The occupancies of the all the located anions were allowed to refine freely and then fixed at the obtained values. Some additional minor occupancy positions of the anions could not be located in the electron density map and were not included in the model resulting in a discrepancy of 2.5 anions per Ag12L6 assembly; these anions are given as PF6 − in the formula given above. Some lower occupancy disordered atoms were modelled with isotropic thermal parameters bond length and thermal parameter restraints were applied to facilitate realistic modelling of the disordered anions. Some acetonitrile solvent molecules were also modelled with partial occupancy.
Further reflecting the solvent loss there is a significant amount of void volume in the lattice containing smeared electron density from disordered solvent (and potentially the unresolved anions). Consequently the SQUEEZE 10 function of PLATON 11 was employed to remove the contribution of the electron density associated with this highly disordered solvent which gave a potential solvent accessible void of 7729 Å 3 per unit cell (a total of approximately 2273 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether they were not included in the formula. Consequently, the molecular weight and density given above are likely to be slightly underestimated. The remaining electron density peaks and holes (up to 1.274 or −1.303 e -Å 3 ) are close to the silver centers reflecting absorption effects or a small amount of unresolved disorder.
CheckCIF gives thirteen B level alerts, all resulting from thermal motion and/or unresolved disorder of the anions as described above.

Specific refinement details:
The crystals of [(ClO4)2⸦Ag12L6]ꞏ10NTf2ꞏ2.5MeCN were grown by diffusion of diisopropyl ether into an acetonitrile solution of the complex formed via the procedure described above. The crystals employed immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data. Despite these measures and the use of synchrotron radiation few reflections at greater than 0.95 Å resolution were observed and the data were trimmed accordingly. Nevertheless, the quality of the data is far more than sufficient to establish the connectivity of the structure. The asymmetric unit was found to contain one complete Ag12L6 assembly and associated counterions and solvent molecules.
Due to the limited resolution, bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other and additional restraints (DFIX, FLAT) were applied to some sections of the organic ligands. Thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for silver to facilitate anisotropic stable refinement. Even with these restraints some thermal parameters remain larger than ideal as a consequence of the high level of thermal motion or minor unresolved disorder present throughout the structure.
Both encapsulated ClO4 − anions were modelled as disordered over two locations. These disordered ClO4 − anions were restrained to have an idealised tetrahedral geometry and the disordered oxygen atoms were modelled with isotropic thermal parameters. The other resolved anions within the structure were all identified as triflimide and also show evidence of substantial disorder. Two triflimide anions were modelled as disordered over two locations and all triflimide anions were modelled with partial occupancy. The occupancies of the triflimide anions were allowed to refine freely and then fixed at the obtained values. These anions were modelled with isotropic thermal parameters and bond length and thermal parameter restraints were applied to facilitate realistic modelling of the disordered anions. Some acetonitrile solvent molecules were also modelled with partial occupancy.
A further 8.3 anions per Ag12L6 assembly remain unaccounted for and no satisfactory model for these anions could be obtained despite numerous attempts at modelling, including with rigid bodies. Therefore the SQUEEZE 10 function of PLATON 11 was employed to account for the highly disordered anions and further disordered solvent molecules, which gave a potential solvent accessible void of 7120 Å 3 per unit cell (a total of approximately 2727 electrons). These anions are included as triflimide in the formula given above. Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether they were not included in the formula. Consequently, the molecular weight and density given above are likely to be underestimated.
CheckCIF gives three A and five B level alerts. These alerts (both A and B level) all result from the limited data resolution, thermal motion and/or unresolved disorder of some anions and solvent molecules and the generally high level of thermal motion present throughout the structure as described above.

Specific refinement details:
The crystals of [(SO4H)2⸦Ag12L6]ꞏ10NTf2ꞏ13MeCNꞏ1.5iPr2O were grown by diffusion of diisopropyl ether into an acetonitrile solution of the complex. The crystals employed immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data. Even so data were obtained to 0.77 Å resolution by employing synchrotron radiation. The asymmetric unit was found to contain one half of a Ag12L6 assembly and associated counterions and solvent molecules.
Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other. Thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for silver to facilitate anisotropic stable refinement. One naphthyridine arm was modelled as disordered over two locations with bond length (DFIX) and FLAT restraints were applied to the disordered parts to ensure a stable refinement.
The oxygen atoms of the central HSO4 − anion was modelled as disordered over three locations with refined occupancies of 0.35068/0.31665/0.33267. Soft bond length and angle restraints were applied to the disordered atoms to achieve a reasonable refinement. While the distance between the disordered HSO4 − anion and its symmetry equivalent is consistent with the presence of a hydrogen bonded dimer, 12 the hydrogen atom of the HSO4 − could not be adequately resolved due to the extensive disorder precluding detailed geometric analysis of the hydrogen bonding. One hydrogen atom was placed on one of the disordered locations, based on a peak in the electron density map with a DFIX restraint applied to the OH bond length and a soft restraint applied to the D-A distance; however the position of this hydrogen is subject to a high degree of uncertainty.
The other anions within the structure, all modelled as triflimide, also show evidence of substantial disorder. All five triflimide anions were modelled as disordered over two or three locations. The occupancies of the disordered anions were allowed to refine freely and then S61 fixed at the obtained values. Some additional minor occupancy positions of the anions could not be located in the electron density map and were not included in the model resulting in a discrepancy of 1.1 counterions per asymmetric unit. Some lower occupancy disordered atoms were modelled with isotropic thermal parameters and bond length and thermal parameter restraints were applied to facilitate realistic modelling of the disordered anions. Some acetonitrile solvent molecules were also modelled with as disordered over multiple locations and/or with partial occupancy. The hydrogen atoms of some disordered acetonitrile molecules could not be located in the electron density map and were not included in the model.
Further reflecting the solvent loss there is a significant amount of void volume in the lattice containing smeared electron density from disordered solvent (and potentially the unresolved anions). Consequently the SQUEEZE 10 function of PLATON 11 was employed to remove the contribution of the electron density associated with this highly disordered solvent which gave a potential solvent accessible void of 1132 Å 3 per unit cell (a total of approximately 401 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diisopropyl ether they were not included in the formula. Consequently, the molecular weight and density given above are likely to be slightly underestimated. The remaining electron density peaks and holes (up to 1.565 or −1.990 e -Å 3 ) are close to the silver centers reflecting absorption effects or a small amount of unresolved disorder.
CheckCIF gives six B level alerts, all resulting from acetonitrile solvent molecules for which hydrogens were not modelled as described above.

Specific refinement details:
The crystals of [(EDS 2-) ⸦ Ag12L6]ꞏ10NTf2ꞏ18MeCNꞏ1.5iPr2Oꞏ0.5H2O were grown by diffusion of diisopropyl ether into an acetonitrile solution of the complex formed via the procedure described above. The crystals employed immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to S62 collect data. Even so data were obtained to 0.7 Å resolution by employing synchrotron radiation. The asymmetric unit was found to contain one half of a Ag12L6 assembly and associated counterions and solvent molecules.
Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other. Thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for silver to facilitate anisotropic stable refinement. One naphthyridine arm was modelled as disordered over two locations with bond length (DFIX) and FLAT restraints were applied to the disordered parts to ensure a stable refinement; one of the silver centers coordinated to the disordered ligand arm was also modelled as disordered over two locations.
The carbon atoms of the encapsulated ethanedisulfonate anion were modelled as disordered over a special position with symmetry enforced half occupancy. The other anions within the structure, all modelled as triflimide, also show evidence of substantial disorder. All triflimide anions were modelled as disordered over two or three locations. Some lower occupancy disordered atoms were modelled with isotropic thermal parameters and bond length and thermal parameter restraints were applied to facilitate realistic modelling of the disordered anions. Some acetonitrile solvent molecules were also modelled with as disordered over multiple locations and/or with partial occupancy. The hydrogen atoms of some disordered acetonitrile molecules and the water molecule could not be located in the electron density map and were not included in the model.
CheckCIF gives sixteen B level alerts, all resulting from thermal motion and/or unresolved disorder of the anions and solvent molecules and solvent molecules for which hydrogens were not modelled as described above.

Specific refinement details:
The crystals of [(S2O8)⸦Ag12L6]ꞏ10NTf2ꞏ12MeCN [+ solvent] were grown by diffusion of diisopropyl ether into an acetonitrile solution of the complex formed via the procedure S63 described above. The crystals employed immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data. Even so data were obtained to 0.85 Å resolution by employing synchrotron radiation. The asymmetric unit was found to contain one half of a Ag12L6 assembly and associated counterions and solvent molecules.
Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other. Thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for silver to facilitate anisotropic stable refinement. One napthyridine arm was modelled as disordered over two locations with bond length (DFIX) and FLAT restraints were applied to the disordered parts to ensure a stable refinement.
The central O-O linkage of the encapsulated peroxydisulfate anion was modelled as disordered over a special position with symmetry enforced half occupancy. The remaining oxygen atoms of this peroxydisulfate anion also show evidence of thermal motion or minor unresolved disorder which could not be adequately modelled with discrete atom positions. The other anions within the structure, all modelled as triflimide, also show evidence of substantial disorder. Four triflimide anions were modelled as disordered over two or three locations and the remaining triflimide was modelled with partial occupancy. The occupancies of the disordered anions were allowed to refine freely and then fixed at the obtained values. Some additional minor occupancy positions of the anions could not be located in the electron density map and were not included in the model resulting in a discrepancy of 1.45 counterions per asymmetric unit. Some lower occupancy disordered atoms were modelled with isotropic thermal parameters and bond length and thermal parameter restraints were applied to facilitate realistic modelling of the disordered anions. Some acetonitrile solvent molecules were also modelled with as disordered over multiple locations and/or with partial occupancy. The hydrogen atoms of some disordered acetonitrile molecules could not be located in the electron density map and were not included in the model.
Further reflecting the solvent loss there is a significant amount of void volume in the lattice containing smeared electron density from disordered solvent (and potentially the unresolved anions). Consequently the SQUEEZE 10 function of PLATON 11 was employed to remove the contribution of the electron density associated with this highly disordered solvent which gave a potential solvent accessible void of 4358 Å 3 per unit cell (a total of approximately 1130 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether they were not included in the formula. Consequently, the molecular weight and density given above are likely to be slightly underestimated.
CheckCIF gives one A level and six B level alerts, all resulting from thermal motion and/or unresolved disorder of the anions and solvent molecules.

S9 Adaption of Cage 1 in the Solid State
The following is an example of the method used to investigate the changes in cage 1 in the solid state for a range of different encapsulated anions. Each vertex of the prism was defined by the centroid between its two silver centers. All measurements were the made from these centroids. Figure S60: An example of distance measurements for (ClO4 − )2 ⊂ 1, with centroids as red dots.
In order to measure the dihedral angles, centroids were calculated for each of the individual disilver centers. Centroids were then calculated between each of the three disilver complexes which define the triangular aperture of the structure. The dihedral angle was then measured from opposing disilver centers through these two central centroids.

S65
The results of the measurements and their averaged values are detailed in Table S1. In order to calculate the area of the triangular apertures at the end of the triangular antiprism one of the three internal angles of the triangular apertures were measured from the centroids generated for the disilver vertices. Figure S62: Measurement of the three angles which define one of the triangular faces of (ClO4 − )2 ⊂ 1.
The area was calculated according to the formula where a and b are the lengths two of the faces of the triangular face and C is the angle defined by these two lines. This calculation was carried out for the two triangular faces of the prism. S66   Table S2: Calculation of the average area of the triangular apertures of (ClO4 − )2 ⊂ 1.
Measurements for the four other crystallogrphic datasets where calculated in a similar manner. The results of these calculates are tabulated in Table S3. Table S3: Calculated average lengths, torsion angles and area of the triangular apertures for cages (X)n ⊂ 1.