Incorporation of a Phosphino(pyridine) Subcomponent Enables the Formation of Cages with Homobimetallic and Heterobimetallic Vertices.

Biological systems employ multimetallic assemblies to achieve a range of functions. Here we demonstrate the preparation of metal-organic cages that contain either homobimetallic or heterobimetallic vertices. These vertices are constructed using 2-formyl-6-diphenylphosphinopyridine, which forms ligands that readily bridge between a pair of metal centers, thus enforcing the formation of bimetallic coordination motifs. Two pseudo-octahedral homometallic MI12L4 cages (MI = CuI or AgI) were prepared, with a head-to-head configuration of their vertices confirmed by X-ray crystallography and multinuclear NMR for AgI. The phosphino-pyridine subcomponent also enabled the formation of a class of octanuclear CdII4CuI4L4 tetrahedral cages, representing an initial example of self-assembled cages containing well-defined heterobimetallic vertices.

Coordination-driven self-assembly provides a powerful tool for the preparation of intricate and functional architectures with relative synthetic ease.[1] The combination of metal ions with well-defined stereoelectronic preferences and ligands that have a rigid arrangement of binding sites has enabled the rational design of polyhedral cage architectures including tetrahedra,[2] cubes,[3] octahedra,[4] and higher-order structures.[5] These cages have attracted considerable interest due to their ability to bind guests within well-defined inner cavities, within which the chemical reactivity and dynamics of guest molecules may be altered.[6]

Most metal–organic cages contain monometallic vertices, as the design principles for these vertices are relatively well-understood. Increased structural complexity and diversity are enabled by the presence of vertices formed from bimetallic units[7] or more complex clusters.[8] Such vertices can also increase the functional complexity, because multiple metal ions can bring about new reactivity.[9] Heterometallic structures[10] are challenging to synthesize in a controlled manner, requiring strategies that include the incorporation of preformed kinetically inert metal–organic building blocks,[11] the use of a mixture of hard and soft ligands that bind different metals preferentially,[12] or the use of ligands with different denticities.[13]

Recently we explored the use of 2-formyl-1,8-napthyridine to prepare cages incorporating disilver vertices.[14] Herein we employ 2-formyl-6-diphenylphosphinopyridine A, a subcomponent containing both N and P donors with nonconverging coordination vectors, as a general method for the construction of metal–organic cages having either homobimetallic or heterobimetallic vertices. Subcomponent A was previously incorporated into a dicopper(I) motif,[15] which was integrated into extended architectures when flexible dianilines were used in combination with rigid carboxylate templates. We reasoned that the combination of A with a more rigid, tritopic aniline would enable the synthesis of more complex metal–organic cages, where the dicopper(I) motif would bring together two aniline residues at the vertices of the cage, without requiring carboxylate templation.

The reaction of A (12 equiv), tris(4-aminophenyl)amine B (4 equiv), and [CuI(MeCN)4](OTf) (12 equiv, –OTf = trifluoromethanesulfonate, triflate)[16] led to the formation of CuI12L4 cage 1 (Figure a), the composition of which was confirmed by ESI-MS. The 1H NMR spectrum of 1 indicated the formation of a high-symmetry product in solution, with the ligand in an environment having 3-fold symmetry. 1H DOSY NMR further confirmed that the aromatic signals corresponded to a single species (Figure b).

Figure 1

(a) Subcomponent self-assembly of MI12L4 cages 1 and 2. Externally coordinated acetonitrile molecules are omitted for clarity. (b) 1H and DOSY NMR spectra of 1. The signal for H6 is not observed at 298 K (see Supporting Information Section 1.2).[19] (c) Crystal structure of 1 with inset showing one of its dicopper(I) vertices. Disorder, anions, solvent of crystallization, and hydrogen atoms are omitted for clarity. (d) Illustration of the head-to-head vertex geometry observed for 1 and the alternative head-to-tail arrangement previously observed in analogous dinuclear CuI complexes.[15]

The crystal structure of 1 revealed a pseudo-octahedral geometry, with a pair of CuI ions occupying each vertex (Figure c). Four faces of the octahedron are occupied by tritopic ligands, while the remaining faces are vacant.[4] Each dimetallic vertex has the same P or M helical twist, with the assembly expressing approximate T point symmetry, consistent with the solution NMR spectra. Both cage enantiomers were observed in the crystal.[17]

The bimetallic vertices display a head-to-head configuration, rather than adopting the head-to-tail arrangement observed in other structures incorporating A(15) and related dicopper(I) complexes[18] (Figure b and d). The internal CuI ion of each vertex is thus chelated by two pyridyl-imine units, and the outer CuI ions are coordinated by two phosphine donors with a further two external acetonitrile molecules completing their tetrahedral coordination spheres. We infer that this arrangement is more favorable than a counterfactual structure with head-to-tail vertices, where the additional acetonitrile ligands would be left inside the cavity to engender steric crowding (see Supporting Information Section 3).

The coordination environments of the inner CuI ions are distorted from a regular tetrahedral geometry, with angles of 66.2–70.4° between the two pyridyl-imine chelate planes and N–CuI–N angles in the range 80.3–139.5°. The outer CuI ions display a more regular tetrahedral geometry, with angles of 97.9–116.9° between ligands. The metal centers of each vertex are separated by 4.02–4.18 Å (average = 4.10 Å), which is much greater than twice the copper(I) van der Waals radius of 1.40 Å,[20] indicating the absence of CuI···CuI interactions.

The inner CuI ions form a regular octahedral framework with an average distance of 12.0 Å along the edges and 16.9 Å between antipodal CuI ions. The cavity of 1 encapsulates a single acetonitrile molecule in the solid state. Its volume was calculated to be 90 Å3 using Molovol.[21]

We reasoned that silver(I) might also form pseudo-octahedral assemblies analogous to 1, as AgI and CuI have similar coordination preferences.[22] Furthermore, 109Ag NMR spectroscopy[23] provides a complementary means to characterize coordination complexes incorporating diamagnetic AgI in solution.[24] Silver(I) complex 2 was thus formed by treating triamine B (4 equiv) with A (12 equiv) and AgIOTf (12 equiv) (Figure a). Its AgI12L4 composition was confirmed by ESI-MS, and its 1H NMR spectrum (Figure c) was again consistent with a high-symmetry structure in solution.

Figure 2

(a) Crystal structure of 2. Disorder, anions, solvent of crystallization, and hydrogen atoms are omitted for clarity. (b) View of one of the disilver vertices of 2, with the imine hydrogens shown as red spheres. The observed 1H–Ag and 31P–Ag couplings are highlighted by red and green arrows, respectively. (c) 1H–109Ag HMBC of 2, revealing a correlation between the imine resonance and the inner silver ions, which resonate at 544 ppm. (d) Coalescence of the imine signal in the 1H NMR spectrum of 2 upon irradiation of 109Ag at 544 ppm. (e) 31P NMR spectra of 2 before and after irradiation of 109Ag at 544 or 950 ppm, resulting in decoupling to the inner and outer 109Ag, respectively.

The crystal structure of 2 confirmed the presence of a pseudo-octahedral assembly (Figure a), analogous to 1, this time with crystallographic T-symmetry. The metal–metal separation at each disilver(I) vertex was found to be 3.38 Å, significantly shorter than the average metal–metal distance of 4.10 Å observed for 1 and slightly greater than twice the van der Waals radius of AgI (1.66 Å).[20] The inner AgI ions form a perfect octahedron with 12.2 Å edges and a distance of 17.3 Å between opposing vertices. The cavity of 69 Å3 (calculated with Molovol[21]) is slightly smaller than that of 1, reflecting a more compressed structure.

The inner AgI ions, once more coordinated by two pyridyl-imine units, are even more distorted from regular tetrahedral geometry (62.6° between pyridyl-imine chelate planes and N–AgI–N angles of 71.1–157.8°) relative to the inner CuI ions of 1, consistent with the greater flexibility of the coordination sphere of silver(I).[25] The outer AgI ion of each vertex is coordinated by a single acetonitrile molecule in an approximately trigonal planar coordination geometry (Figure b). The coordinated acetonitriles were not observed by 1H NMR, presumably due to rapid exchange with CD3CN.

The solution structure of 2 was further probed through multinuclear NMR experiments (Figure c–e), which confirmed the presence of two distinct AgI environments, corresponding to the inner and outer silver ions at each vertex. These data indicate that the solution structure mirrors the solid-state one. The imine signal in the 1H NMR spectrum of 2 split into a doublet (Figure d), in contrast to the singlet observed for 1. In the case of 2, coupling arises between the imine proton and the nearby internal AgI ion with a 109Ag chemical shift of 544 pm, as determined from a 1H–109Ag HMBC spectrum (Figure c).[26]

The 31P NMR spectra of 2 (Figure e) showed complex splitting patterns, consistent with coupling between the phosphine and both unique AgI ions. A major coupling was observed to the external AgI ions, with further fine splitting resulting from longer-range coupling to the internal AgI ion, which partially collapsed upon irradiation of the inner 109Ag resonance at 544 ppm. Stimulation of 109Ag over a broad window in approximately 50 ppm increments (Figure S25) allowed identification of a resonance at ca. 950 ppm, corresponding to the outer AgI ions.

Because structures 1 and 2 possess two distinct coordination environments, we hypothesized that subcomponent A might also be capable of stabilizing assemblies with heterobimetallic vertices. We initially investigated whether CuI and AgI could be selectively incorporated into the two distinct binding sites at the vertices of the pseudo-octahedral framework shared by 1 and 2. However, the reaction of trianiline B (4 equiv) and A (12 equiv) with equimolar amounts of [CuI(MeCN)4](OTf) and AgIOTf (6 equiv each) led to the formation of a distribution of CuIAgI(12–L4 pseudo-octahedral species (Figure S26). We infer that the similarity in coordinative preferences between CuI and AgI led to the formation of these mixed-metal species.

We hypothesized that a metal ion with different coordinative preferences, such as cadmium(II), would lead to discrimination between the different binding sites when combined with copper(I). The self-assembly of triamine B (4 equiv) and A (12 equiv) with [CuI(MeCN)4](ClO4) (4 equiv) and CdII(ClO4)2 (4 equiv) gave rise to a new product (3), which displayed a single 1H NMR signal for each type of ligand proton (Figure a).[27] ESI-MS revealed a CdII4CuI4L4 composition, distinct from pseudo-octahedral assemblies 1 and 2.

Figure 3

Subcomponent self-assembly of CdII4CuI4L4 cages (a) 3 and (b) 4. Externally coordinated acetonitrile molecules are omitted for clarity. (c) Crystal structure of 3 with inset showing one CdIICuI vertex. (d) Crystal structure of 4. Disorder, anions, solvent of crystallization, and hydrogen atoms are omitted for clarity.

Single-crystal X-ray analysis confirmed the face-capped tetrahedral structure of 3 (Figure c). The heterobimetallic vertices of 3 each consist of an inner CdII and an outer CuI, separated by distances of 3.47–3.57 Å (average 3.52 Å), greater than the sum of the van der Waals radii of the two ions (2.98 Å).[20] This vertex geometry enables aromatic stacking to occur between a phosphorus-bound phenyl ring from each ligand and the pyridine of a neighboring ligand, with distances of 3.1–3.4 Å between stacked rings (Figure c, inset). Such stacking was not observed in the homobimetallic vertices of 1 and 2.

The CdII ions bring together three pyridyl-imine ligands at each vertex. The resulting coordination geometry is flattened from a regular octahedral arrangement, with N–CdII–N angles of 71.5–112.0° between cis-coordinated nitrogen donors. The CuI ions are coordinated by a phosphine donor from each ligand, with a single acetonitrile molecule completing the tetrahedral coordination sphere.

Coordination of CdII to the pyridyl-imine donors within 3 allows them to adopt their preferred six-coordinate configuration, leaving the phosphine donors free to bind CuI in an approximately tetrahedral configuration. Although both metal ions are classed as soft acids, the lower charge of CuI renders it softer than CdII, and thus with a greater propensity to coordinate to the softer phosphine donors.[28]

The structure of 3 evokes previously reported MII4L4 tetrahedra,[2a,29] with all octahedral CdII ions within each cage sharing the same Δ or Λ stereochemistry, and the face-capping ligands also adopting a propeller-like helical arrangement. The CdII ions are separated by an average distance of 12.6 Å. A cavity volume of 51 Å3 was calculated using Molovol,[21] within the range observed for analogous tetrahedral cages assembled from B, 2-formylpyridine, and FeII or CoII (31 and 63 Å3 respectively, calculated using the same method).[2a,29] The central nitrogen atoms of each ligand are slightly pyramidalized to point outward, with C–N–C angles ranging from 115.1° to 118.1° (average 117.3°). This observation contrasts with the structures of 1 and 2, where the central nitrogen atoms are nearly planar, with average C–N–C angles of 119° and 120°, respectively.

To investigate the generality of this approach for forming heterometallic cages, we also prepared a larger tetrahedral cage based on triamine C, which was shown to produce MII4L4 tetrahedra with rich host–guest chemistry.[2b] Treatment of subcomponents C (4 equiv) and A (12 equiv) with [CuI(MeCN)4](ClO4) (4 equiv) and CdII(ClO4)2 (4 equiv) yielded CdII4CuI4L4 structure 4 (Figure b), as confirmed by ESI-MS. 1H NMR spectra were again consistent with a T-symmetric structure in solution.

The crystal structure of 4 confirmed the formation of a face-capped tetrahedral cage with heterobimetallic CdIICuI vertices, similar to those of 3 (Figure d). The internal CdII cations are separated from one another by an average distance of 16.3 Å, greater than in 3, and the 240 Å3 cavity of 4 is also correspondingly larger, calculated using Molovol.[21] Future work will compare the guest encapsulation abilities of this cavity with that of the analogous MII4L4 tetrahedron.

Subcomponent A represents a rare example of a building block that can generate either homobimetallic or heterobimetallic coordination motifs, resulting in two structurally distinct families of coordination cages. The two chemically distinct coordination environments formed from the previously unreported head-to-head arrangement of A have enabled access to cages with heterobimetallic vertices for the first time. Future work will investigate whether the labile coordination sites of the cages, occupied by acetonitrile molecules in the solid state, could enable further functionalization of the cage exterior, to allow tuning of their solubility,[30] the attachment of fluorescent tags for biomedical applications,[31] or the chirality of the cages to be controlled.[32] Explorations may also be fruitful of the mutual influences of the two vertex metal ions, bound at well-defined distances from each other, on the electrochemical properties of the cages and their host–guest properties. Future studies will also seek to exploit the potential photophysical properties[33] of the copper(I)-based cages reported herein for sensing or optoelectronic applications.[34]