Self-assembly of nanoscopic coordination cages of D_3h symmetry

Abstract

A family of nanoscale-sized supramolecular cage compounds with a trigonal prismatic framework was prepared by means of spontaneous self-assembly from the combination of a predesigned molecular “clip” with tritopic pyridyl subunits. As confirmed by x-ray crystallography, the smallest structure of the reported series is ≈1 × 2 nm and possesses a nitrate anion incarcerated inside its molecular cavity. The largest structure has dimensions of ≈ 1 × 4 nm.

The formation of discrete supramolecular entities driven and held together by metal coordination is an intense new area of investigation at the forefront of supramolecular chemistry (1–10). Because self-assembly is guided by the chemical information encoded into the molecular subunits, diverse structures with predetermined shape, size, and functionality can readily be designed. Indeed, a wide variety of aesthetic structures have been realized, such as molecular grids, helicates, rings, catenanes, tetrahedra, cubes, cuboctahedra, etc. Once assembled, many of the hollow structures have been shown to be capable of encapsulating molecules through electrostatic and/or dispersion forces. Often times, ions will template the formation of an assembly (11–21). When considering that metal-containing assemblies often possess magnetic, photophysical, and/or redox properties not accessible from purely organic systems, studies in basic host–guest chemistry hold new promise for technologies in molecular sensing (22–28), separations, and catalysis (29, 30).

Because lower-symmetry hosts can ultimately be expected to show enhanced guest selectivity, especially toward planar aromatic guests, prismatic cages represent a natural progression in the development of this area. Although M₃L₂-type cages are relatively simple three-dimensional constructs, they remain uncommon. Of those that have been reported (31–40), most usually either require the use of templates to assemble in solution, or assemble only in the solid state. Part of the reason for this limitation is possibly the fact that, in most cases, flexible ligands were used. By contrast, structures derived from rigid tritopic linkers with cis-metal ions are either: (i) tetrahedral M₆L₄ cages (41) where L is a planar ligand, or (ii) double-square M₆L₄ cages (42) where L is a 109° linker ligand. Construction of the M₃L₂, D_3h species is complicated by the fact that rigid tritopic linkers with ideal mutual angles of 60° are not easily accessible. A noteworthy trigonal bipyramidal structure (35), made from Pd(II) ions and a calix[3]arene subunit, was shown to be able to reversibly bind a molecule of C₆₀.

By exploiting incommensurate symmetry requirements for differing metallic subunits, an alternative approach to structures of this general topology was recently reported. Raymond and Wong (43–45) successfully prepared a series of M₂M′₃L₆ supramolecular clusters where a multifunctionalized ligand (L) was cleverly designed to selectively interact with two types of metal ions (one hard and one soft).

Double oxidative addition to rigid aromatic ligands offers a unique way to direct the open coordination sites of metals (46), because nearly any desired angle can be fashioned provided the corresponding aryl halide is accessible. Presented here is a ligand-directed approach to molecular prisms, thus providing a versatile method for the construction of supramolecular structures of variable size and functionality, in high yield, and without the use of templates. Specifically, the modular self-assembly of a new family of metalla-bicyclic structures (3a–3c), directed by a molecular “clip” (1) as a preconstructed shape-defining unit, is described. The structure and properties of these nanoscopic, macrocyclic species have been studied by electrospray ionization (ESI)-mass spectrometry, multinuclear NMR, and x-ray crystallography.

Experimental Section

General Methods.

Building unit 2a (47) was synthesized by a published method and 2b and 2c were prepared in an analogous manner. NMR spectra were recorded on a Varian XL-300 or a Unity 300 spectrometer. Deuterated solvents were used as received. Proton chemical shifts are reported relative to residual protons of the deuterated acetone (δ 2.05). ³¹P {¹H} chemical shifts (δ) are reported relative to an unlocked, external sample of H₃PO₄ (0.0 ppm). Elemental analyses were performed by Atlantic Microlab, Norcross, GA.

General Procedure for the Preparation of Compounds 3a–3c.

In a 2-dram (8 g) vial equipped with a magnetic stir bar were placed solid 1,8-bis(trans-Pt(PEt₃)₂(NO₃))anthracene (20.0 mg, 0.0172 mmol) with a 0.67 molar ratio of the appropriate tritopic bridging ligand. Next, 1 ml of an acetone-d₆/D₂O mixture was added to the vial (1:1 for 3a, 1:2 for 3b and 3c), which was then sealed with Teflon tape and heated in an oil bath at 50–55°C, with stirring. After the specified length of time (3–5 h for 3a, 20–24 h for 3b and 3c), the initial pale yellow suspension gradually became homogeneous. The solution was then transferred to an NMR tube for analysis. The product was precipitated with KPF₆, collected on a frit, washed with excess water, and dried in vacuo.

Bicyclo[Tris[1,8-bis(trans-Pt(PEt₃)₂(NO₃))anthracene)]bis[Tris(4-pyridyl)methanol]] (3a).

¹H NMR (acetone-d₆/D₂O, 300 MHz): δ 9.18 (d, 6H, ³J_HH = 6.3 Hz, H_α-Py), 8.98 (m, 9H, H_α-Py, H₉), 8.37 (s, 3H, H₁₀), 8.31 (d, 6H, ³J_HH = 6.0 Hz, H_β-Py), 7.75 (m, 12H, H_β-Py, H_{4, 5}), 7.69 (d, 6H, ³J_HH = 7.8 Hz, H_2,7), 7.14 (m, 6H, H_3,6), 1.40 (m, 72H, PCH₂CH₃), 0.89 (m, 108H, PCH₂CH₃). ³¹P{¹H} NMR (acetone-d₆/D₂O, 121.4 MHz): δ 9.5 (s, ¹J_PPt = 2,652 Hz). ESI-MS: 2,069 [calculated for (M − 2PF₆)²⁺: 2,069], 1,331 [calculated for (M − 3PF₆)³⁺: 1,331]. Analysis calculated for C₁₉₆H₂₃₀F₃₀N₇O₅P₁₇Pt₆: C, 39.58; H, 5.23; N, 2.21. Found: C, 39.64; H, 5.24; N, 2.21.

Bicyclo[Tris[1,8-bis(trans-Pt(PEt₃)₂(NO₃))anthracene]bis[1,3,5-Tris(4- ethynylpyridyl)adamantane]] (3b).

¹H NMR (acetone-d₆/D₂O, 300 MHz): δ 9.17 (s, 3H, H₉), 8.95 (d, 6H, ³J_HH = 5.7 Hz, H_α-Py), 8.76 (d, 6H, ³J_HH = 5.7 Hz, H_α-Py), 8.32 (s, 3H, H₁₀), 7.85 (m, 12H, H_β-Py), 7.66 (m, 12H, H_2,4,5,7), 7.13 (m, 6H, H_3,6), 1.37 (m, 72H, PCH₂CH₃), 0.82 (m, 108H, PCH₂CH₃). ³¹P{¹H} NMR (acetone-d₆/D₂O, 121.4 MHz): δ 8.9 (s, ¹J_PPt = 2,656 Hz). ESI-MS: 2287 [calculated for (M − 2PF₆)²⁺: 2,287], 1,476 [calculated for (M − 3PF₆)³⁺: 1,476), 1,071 [calculated for (M − 4PF₆)⁴⁺: 1,071). Analysis calculated for C₂₀₄H₂₇₂F₃₆N₆P₁₈Pt₆: C, 46.94; H, 5.25; N, 1.61. Found: C, 47.07; H, 5.27; N, 1.71.

Bicyclo[Tris[1,8-bis(trans-Pt(PEt₃)₂(NO₃))anthracene]bis[1,1,1-Tris(4- phenyl(4′-ethynylpyridyl)ethane]] (3c).

¹H NMR (acetone-d₆/D₂O, 300 MHz): δ 9.33 (s, 3H, H₉), 9.06 (d, 6H, ³J_HH = 5.4 Hz, H_α-Py), 8.91 (d, 6H, ³J_HH = 5.7 Hz, H_α-Py), 8.40 (s, 3H, H₁₀), 8.03 (m, 12H, H_β-Py), 7.73 (d, 6H, ³J_HH = 8.7 Hz, H_4,5), 7.71 (d, 6H, ³J_HH = 6.0 Hz, H_2,7), 7.62 (d, 12H, ³J_HH = 8.4 Hz, H_{3,5-phenylene}), 7.24 (d, 12H, ³J_HH = 8.7 Hz, H_{2,6-phenylene}), 7.17 (m, 6H, H_3,6), 1.46 (m, 72H, PCH₂CH₃), 0.87 (m, 108H, PCH₂CH₃). ³¹P{¹H} NMR (acetone-d₆/D₂O, 121.4 MHz): δ 8.8 (s, ¹J_PPt = 2,641 Hz). ESI-MS: 2409 [calculated for (M − 2PF₆)²⁺: 2,409], 1,557 [calculated for (M − 3PF₆)³⁺: 1557), 1,132 [calculated for (M − 4PF₆)⁴⁺: 1,132]. Analysis calculated for C₁₉₆H₂₅₈F₃₆N₆P₁₈Pt₆: C, 46.07; H, 5.09; N, 1.64. Found: C, 45.26; H, 5.03; N, 1.74.

X-Ray Data Collection, Structure Solution, and Refinement.

A single crystal of either 3a or 3c was selected from the reaction product and mounted on a thin glass fiber by using silicone grease. The data were collected at −100°C, using a narrow frame method with scan widths of 0.3° in ω and exposure times of 20 s. A hemisphere of intensity data were collected in 1,081 frames with the crystal-to-detector distance of 50.4 mm, which corresponds to a maximum 2θ value of 54.1°. Frames were integrated with the Bruker SAINT program (Billerica, MA). A semiempirical absorption correction based on simulated ψ-scans was applied to the data set. Both structures were solved by a combination of direct methods and difference Fourier methods, and refined with full-matrix least squares techniques. Details of the data collection, structure solution, and refinement are given in Table 1. Tables of the atomic coordinates, thermal displacement parameters, and selected bond distances and angles complete with ORTEP representations of 3a and 3c are provided in the supporting information, which is published on the PNAS web site, www.pnas.org.

Table 1.

Crystal data and structure refinement results for 3a and 3c

Empirical formula	C146H230F36N6O2P18Pt6 (3a)	C194H258F36N6P18Pt6 (3c)
Formula weight	4,513.36	5,086.06
Color/shape	Yellow/trigonal prism
Crystal system	Triclinic	Rhombohedral
Space group	P1̄ (#2)	R3̄c (#167)
Unit cell dimensions	a = 20.6055(5) Å	a = 30.6988(2) Å
	b = 20.8610(5) Å	b = 30.6988(2) Å
	c = 29.2406(5) Å	c = 52.3624(8) Å
	α = 79.88(1)°	α = 90.00°
	β = 69.55(1)°	β = 90.00°
	γ = 61.39(1)°	γ = 120.00°
Volume	10,337.9(4) Å3	42,735.9(8)
Z	2	6
Density (calculated)	1.450 mg/m3	1.186 mg/m3
Absorption coefficient	4.257 mm−1	3.096 mm−1
F(000)	4,452	15,156
Crystal size	0.31 × 0.19 × 0.17 mm	0.24 × 0.17 × 0.15
θ Range for data collection	1.49 to 28.40	1.72 to 28.21
Reflections collected	123,286	11,538
Independent/observed reflections	48,757 (Rint = 0.088)/23700 ([I > 2σ(I)])	88,416 (Rint = 0.1265)/4256 ([I > 2σ(I)])
Absorption correction	Semiempirical from simulated Ψ-scans	Semiempirical from simulated Ψ-scans
Range of relative transmission factors	1.00, 0.841	1.00, 0.883
Final R indices [I > 2σ(I)]	R = 0.0634, Rw = 0.1688	R = 0.1492, Rw = 0.3827
Goodness-of-fit on F	0.965	1.300
Weighting scheme	1/[σ2(F) + (0.1064P)2 + 0.0000P] where P = (F + 2F)/3w	1/[σ2(F) + (0.2000P)2 + 0.0000P], where P = (F + 2F)/3w
Largest difference peak and hole	3.47 and −2.33 eÅ−3	3.77 and −1.08 eÅ−3

Results and Discussion

The preparation of several discrete nanoscopic cage complexes can be achieved with simplicity and, given the proper conditions, in essentially quantitative yield. The assembly of these cages is most easily monitored by their ¹H and ³¹P{¹H} NMR spectra. For example, on gentle heating, a suspension of a 3:2 stoichiometric combination of 1 and 2a in a 1:1 (vol:vol) acetone-d₆/D₂O mixture (Scheme S1) gradually dissolves (3 h) to give a bright yellow homogeneous solution of the smallest molecular cage assembly described in this paper, structural motif 3a. ³¹P{¹H} NMR (121.4 MHz) analysis of the reaction solution is consistent with the formation of a highly symmetrical species by the appearance of a sharp singlet with concomitant ¹⁹⁵Pt satellites, shifted 5.0 ppm upfield (−Δδ) relative to 1 (Δ¹J_PPt = −203 ppm). Likewise, reaction of component 1 with tritopic bridging ligands 2b and 2c yielded the analogous molecular cages 3b and 3c, respectively, of differing dimensions and topology (Scheme S1).

Scheme 1.

Self-assembly of coordination cages 3a–3c.

Examination of the ¹H NMR (300 MHz) spectra of cages 3a–3c was also indicative of highly symmetrical structures and displayed significant spectroscopic differences from their monomeric subunits. Particularly diagnostic were the significant downfield shifts of the pyridyl signals (Δδ ≈ 0.5 ppm), associated with the loss in electron density on coordination by the nitrogen lone pair to the platinum metal center. In accordance with previously reported molecular rectangles (48, 49), the inner and outer pyridyl protons were found to be inequivalent because of restricted rotation around the platinum–pyridine bond (50–54). After the reactions were judged to be complete by NMR, each product was isolated as its hexafluorophosphate salt by precipitation with excess KPF₆.

Additional support for these structures was provided by ESI-MS. Analysis of the hexafluorophosphate salts of 3a–3c all showed prominent M − 2PF₆ and M − 3PF₆ peaks (isotopically unresolved). Interestingly, the respective 2⁺ and 3⁺ peaks for 3a at m/z 2,069 and 1,331 were found to be consistent with only five of the six nitrate counterions exchanged for PF. This formulation was also supported by elemental analysis. The nature of this distinction was revealed by single crystal x-ray analysis of 3a.

Quality single crystals of 3a grew after 2 days at ambient temperature by vapor diffusion of diethyl ether into a concentrated acetonitrile solution of the complex. It crystallizes in the triclinic system (space group P) with an asymmetric unit cell that contains a whole molecular prism, an NO and five PF anions. The entire molecular prism is situated so as to void any crystallographically imposed symmetry on the molecule. However, a pseudo-3-fold rotation axis can still be identified to run through the two OH groups. The most significant feature is that a nitrate anion was found incarcerated inside the cavity of the cationic complex (Fig. 1). It appears to be strongly bound because it was not exchanged, even in the presence of a large excess of KPF₆. It is interesting to note that the principal molecular axis of the NO is approximately in alignment with the pseudo-C₃ axis of the prism; the three oxygen atoms of the nitrate anion extrude into the space created between the molecular “clips.” It is tempting to conjecture that the size and symmetry match between the supermolecular cation and the nitrate anion may contribute a great deal to this cation–anion affinity. As a point of note, the assembly of 3a was observed to be about an order of magnitude faster than the formation of 3b and 3c. This finding suggests that the nitrate anion is possibly providing a templating effect, facilitating the formation of 3a. Computer-generated van der Waals surfaces based on the x-ray data show that the NO fits securely inside the host framework. The solid-state asymmetry of the molecule is probably caused by an outward distortion of the Pt–N bonds that appear to be a result of slightly mismatched angles between the rigid molecular subunits; the average trans-(C—Pt—N) bond angle is about 174°. The dimensions of 3a offer a more detailed description of the overall topology. The diameter is ≈19.1 Å, as defined by a circle that the arms of the propeller sweep out when rotated on its 3-fold axis (measured to the centroids of the anthracenes). The height, as defined by the distance between the central carbon atoms of ligand 3a, is 9.34 Å. As expected, the pyridine groups were found perpendicular to the coordination planes of the platinum metal centers.

Figure 1.

Two perspectives of the x-ray structure of 3a. (Upper) The framework is shown as a stick model, and the encapsulated nitrate is represented with a space-filling model. All other counterions and the methyl groups from the triethylphosphines are omitted for clarity. (Lower) Space-filling model viewed down the C₂ axis. Colors: C, blue; H, white; N, green; O, red; P, orange; Pt, gray.

Single crystals of the BF salt of 3b were grown from a DMF/Et₂O solvent system. Examination and geometric data collection of three single crystals from different batches of samples consistently revealed that 3b crystallizes in the C-centered orthorhombic system with a = 54.868(4) Å, b = 76.086(4) Å, c = 27.958(2) Å, and V = 116714(1) Å3. Intensity data collection on a yellow-orange single crystal with dimensions of 0.33 × 0.25 × 0.21 mm at −100°C afforded a total of 483,454 reflections. The systemic absence of the reflections suggested the possible space group to be C222₁. Unfortunately, neither Patterson approaches nor direct methods have yielded any reasonable structure solution thus far. However, given that the NMR spectra and the ESI-MS data for 3b are consistent with the smaller prism (3a, see above), as well as the larger prism (3c, see below), the structures of which have both been unambiguously determined by x-ray crystallography, the structure of 3b can be assigned as a trigonal prism with a high degree of confidence.

Quality single crystals of the largest assembly in this series, metalla-prismatic cage 3c, grew over the course of 1 week at ambient temperature by vapor diffusion of diethyl ether into a concentrated DMF solution of the complex. The asymmetric unit cell of 3c (space group Rc) contains half of a molecular “clip” (1) and a third of ligand 2c. The entire molecular trigonal prism is generated by means of a 3-fold rotary inversion axis that runs through the central carbon atoms of the ligands (Fig. 2). Therefore, the molecule possesses crystallographically imposed D_3h symmetry. Because of the domination by the heavy Pt atoms on the electron density maps, three terminal carbon atoms on the triethylphosphine ligands could not be located from the difference Fourier maps. On the other hand, the PF ions are located completely outside the prism, and situated near a 2-fold rotation axis, making it crystallographically disordered. The structure refinement of 3c was based on a model reflecting such disorders and gave r = 0.149. The packing diagram shows that the molecular prisms stack on top of each other along the crystallographic {001} direction (Fig. 3). Because the molecular axes of the prisms coincide with each other and the molecular clips from the neighboring prisms are in a staggered conformation, propeller-like hexagonal patterns are generated by the cations packed throughout the crystal lattice. The greater flexibility of the arms of the ligand could account for the higher symmetry in the solid-state structure of 3c vs. 3a. This bending is accommodated primarily by the acetylenic moieties. The overall dimensions of this structure are about 38 Å in diameter, with a distance of about 11.5 Å separating the central carbon atoms of ligand 2c. These dimensions place 3c among the largest discrete metallacyclic assemblies characterized by x-ray crystallography to date.

Figure 2.

Two perspectives of the x-ray structure of 3c. (Upper) Stick model as viewed down the C₃ axis. Counterions and the methyl groups from the triethylphosphines are omitted for clarity. (Lower) Space-filling model viewed down the C₂ axis. Colors: C, blue; H, white; N, green; P, orange; Pt, gray.

Figure 3.

Crystal packing diagram of 3c as viewed down the {001} direction.

Conclusions

In summary, metalla-bicyclic supramolecular cages 3a–3c were prepared by means of spontaneous self-assembly based on a molecular “clip” (1) as a rigid modular subunit. Multinuclear NMR and ESI mass spectral data, in conjunction with x-ray crystallography, unambiguously established their structure, which showed that the sizes of the propeller-type three-dimensional structures range from 1 × 2 nm to 1 × 4 nm. The formation of the smallest of these structures, 3a, was accompanied by the entrapment of a nitrate ion inside its molecular cavity. Although the chemistry of metal-assembled molecular containers is steadily expanding, many of the desired features of that comprise guest selectivity are far from fully developed. Anionic guests appear to be the simplest and easiest starting points for controlled studies of inclusion phenomenon because of the inherent electrostatic affinity for the host framework. Ongoing investigations for this and related series of structures are being undertaken to assess the desired features of guest selectivity, reversibility, and dynamics.

Abbreviation

ESI

electrospray ionization