Systematic assembly of the double molecular boxes: {Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃} as a tridentate ligand

Abstract

Cubic cage compounds composed of Co-CN-Ru linkages have been prepared which illustrate the following features: (i) new motifs for alkali metal ion complexation (i.e., cationic receptors for cations), (ii) a new family of triaza-metalloligands, and (iii) a double box-like cage. The cages are synthesized by the condensation of [CpCo(CN)₃]⁻ and [Cp*Ru(NCMe)₃]⁺ (cyclopentadienyl, Cp; pentamethylcyclopentadienyl, Cp*) the presence of Cs⁺. The species {Cs⊂{[CpCo(CN)₃]₄[Cp*Ru]₄}⁺ and {Cs⊂{[CpCo(CN)₃]₄[Cp*Ru]₃[Cp*Rh]}²⁺ illustrate the box-completion reaction Cs ⊂ Co₄Ru₃ + M (M = Cp*Rh²⁺, Cp*Ru⁺). With the naked ion precursors [Na(NCMe)₆]⁺ and [Fe(NCMe)₆]²⁺, the box-completion reactions afforded {Na{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃]}₂}⁺ and {Fe{Cs⊂ [CpCo(CN)₃]₄[Cp*Ru]₃]}₂}²⁺. These cages provide the first examples, to our knowledge, of the double-box motif.

The design of molecular containers represents an important component of nanotechnology and has attracted intense interest from synthetic chemists (1–4). Research on molecular containers can realistically be expected to provide highly selective sensors, sorters, and catalysts for numerous applications. A significant challenge in this area is the development of containers that are stereochemically rigid, because rigidity is the basis of sterically governed selectivity. Rigidity, however, is incompatible with much of organic chemistry, and this dichotomy is problematic because organic (and organometallic) chemistry provides the most versatile construction tools for the synthesis of molecular containers. In this contribution, we address this dichotomy, i.e., the incorporation of organic motifs into rigid frameworks. Our approach involves a hybridization of organometallic chemistry and well established precedents in the chemistry of metal cyanides.

Cyanometallates are metal complexes with the general formula L_lM_m(CN)_n. The most important cyanometallate is Prussian blue (PB), an inorganic polymer with the formula Fe₇(CN)₁₈(H₂O)_x (x ∼ 15) (5). The synthesis of this useful solid arises from the condensation of [Fe(CN)₆]⁴⁻ and Fe(III) salts (Eq. 1).

1

The structure of PB may be roughly described as interconnected cubic cage subunits with Fe vertices linked by cyanide. The PB structure is in fact complicated because the otherwise idealized cubic framework is interrupted by vacancies at the metal positions, these vacancies being occupied by water molecules (6). A building block approach is inherent in Eq. 1, i.e., the use of preassembled [Fe(CN)₆]⁴⁻ units that remain intact throughout their condensation with the labile Fe³⁺ precursors. Recognition of this building block aspect has spawned intense exploratory studies by using variations of the original PB synthesis, especially focused on other polycyanometallates (e.g., [Mo(CN)₇]⁴⁻, [V(CN)₆]³⁻, and [Ni(CN)₄]²⁻) (7–9) and, in place of Fe(III) in Eq. 1, coordination complexes containing some nondisplaceable ligands (e.g., [Ni(H₂NCH₂CH₂NH₂)₂]²⁺ as a bridging, doubly Lewis-acidic metal center) (10). The PB analogues are of continuing interest as sources of molecular magnets, solid sorbents, and electrode materials.

In recent years we have developed families of molecular cyanometallate ensembles that are synthesized analogously to PB, except that our molecular building blocks are tricyanometallates wherein the three cyanide ligands are mutually cis. Half of the coordination sphere of these tricyanometallates is occupied by a strongly coordinating nondisplaceable coligand. The face-capping coligand inhibits the formation of polymers by minimizing crosslinking but still promotes the formation of three-dimensional structures, which resemble subunits of PB. Particularly effective as face-capping ligands are cyclopentadienyl, C₅H₅ (Cp), and its pentamethyl analogue, C₅Me₅ (pentamethylcyclopentadienyl, Cp*). In a proof of concept experiment, we showed that [CpCo(CN)₃]⁻ and [Cp*Rh(MeCN)₃]²⁺ condense with displacement of MeCN to give M₈(CN)₁₂ and related M₇(CN)₁₂ cages, which we refer to as cyanometallate boxes and bowls, respectively (Eqs. 2 and 3; refs. 11 and 12).

2

3

The organometallic approach is powerful because numerous CpM and Cp*M precursor reagents are available, which enable investigations of diverse collections of cages that differ in charge, steric properties, and kinetic lability. Furthermore, one can use ligands in place of C₅R₅, such as 1,4,7-triazacyclononane (13). In this report, we explore the impact of the seemingly subtle replacement of Cp*Rh²⁺ in Co₄Rh₄⁴⁺ and Co₄Rh₃²⁺ (Eqs. 2 and 3) by Cp*Ru⁺. Both Cp*Rh²⁺ and Cp*Ru⁺ are 12e⁻ fragments that are in common use in organometallic chemistry. The structural and host-guest properties of the resulting Co-Ru cages are not only rich but also predictable, and the latter behavior has allowed the synthesis of an example of a molecular double box.

Materials and Methods

General.

Standard Schlenk techniques were used in all syntheses. The precursors [Cp*Rh(NCMe)₃](PF₆)₂ (14) and [Fe(NCMe)₆](PF₆)₂ (15) were prepared according to published procedures. Because the purity of [Cp*Ru(NCMe)₃]PF₆ (16) proved critical, the as-synthesized salt (4) was further recrystallized from CH₂Cl₂-Et₂O. ¹³³Cs NMR spectra were obtained at 78 MHz, and the chemical shift of CsOTf (OTf, trifluoromethanesulfonate) (MeCN solution) was referenced at δ34.362.

{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₄}PF₆⋅2H₂O.

A solution of 102 mg (0.202 mmol) of [Cp*Ru(NCMe)₃]PF₆ in 20 ml of MeCN was added dropwise to a solution of 14 mg (0.051 mmol) of CsOTf and 102 mg (0.202 mmol) of [K(18-crown-6)][CpCo(CN)₃] in 20 ml of MeCN. The resulting red solution was stirred for 1 h, and then solvent was reduced to about 5 ml under vacuum. The concentrate was diluted with 20 ml of Et₂O to precipitate the product as a dark red powder, which was collected by filtration, washed with 5:1 Et₂O/MeCN (vol/vol) and Et₂O, and dried under vacuum. Yield: 96 mg (93%). IR (KBr, cm⁻¹): ν_CN = 2174, 2119. ¹H NMR (CD₃CN): δ1.679 (s, 15H), 5.568 (s, 5H). ¹³³Cs NMR (CD₃CN): δ1.33. Electrospray ionization (ESI)-MS (m/z): 1887 ({Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₄}⁺). Anal. Calcd for C₇₂H₈₄Co₄CsF₆N₁₂O₂PRu (Found): C, 41.83 (41.98); H, 4.27 (4.10); N, 8.13 (8.02).

{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}.

A solution of 100 mg (0.198 mmol) of [Cp*Ru(NCMe)₃]PF₆ in 15 ml MeCN was added dropwise to a solution of 133.6 mg (0.264 mmol) of [K(18-crown-6)][CpCo(CN)₃] and 18.6 mg (0.066 mmol) of CsOTf in 5 ml of MeCN. Immediately, the solution became deep red, and a violet-brown solid precipitated over the course of 30 min. The solid was collected, washed with Et₂O, and recrystallized from 5 ml of CH₂Cl₂ by the addition of 20 ml of Et₂O. Yield of violet-brown microcrystals: 86 mg (79%). (Note: Cs⊂Co₄Ru₃ is soluble in tetrahydrofuran (THF), insoluble in MeCN, and unstable in CH₂Cl₂.) ¹H NMR (δ, THF-d₈) indicates a mixture of two isomers in a 4:1 ratio. Major isomer: 1.663 (s, 30H), 1.689 (s, 15H), 5.430 (s, 10 H), 5.452 (s, 5H), 5.593 (s, 5H); minor isomer: 1.636 (s, 15H), 1.679 (s, 30H), 5.384 (s, 10 H), 5.505 (s, 5H), 5.624 (s, 5H). 78 MHz ¹³³Cs NMR (THF): δ41.462 (major isomer), δ89.321 (minor isomer). IR (KBr): 2163, 2124 cm⁻¹. ESI-MS: m/z = 1649 (Cs[CpCo(CN)₃]₄[Cp*Ru]₃). Anal. Calcd for C₆₂H₆₅Co₄CsN₁₂Ru₃⋅4.5CH₂Cl₂ (Found): C, 39.30 (39.23); H, 3.67 (3.65), N, 8.27 (8.27).

{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃[Cp*Rh]}(PF₆)₂.

A solution of 19 mg (0.03 mmol) of [Cp*Rh(NCMe)₃](PF₆)₂ in 25 ml of 5:1 CH₂Cl₂:MeCN was added dropwise to a solution of 49 mg (0.03 mmol) of Cs⊂Co₄Ru₃ in 20 ml of CH₂Cl₂. The resulting solution was stirred for 1 h, allowed to stand for 24 h, and finally diluted with 60 ml of Et₂O to precipitate a dark red powder, which was collected by filtration, washed with THF and Et₂O, and dried under vacuum. Yield: 43 mg (66%). ¹H NMR (CD₃CN): δ1.675 (s, 45H), 1.805 (s, 15H), 5.616 (s, 5H), and 5.687 (s, 15H). 78 MHz ¹³³Cs NMR (CD₃CN): δ10.225. ESI-MS (m/z): 944.4 ({Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃[Cp*Rh]}²⁺). Anal. Calcd for C₇₂H₈₀Co₄CsF₁₂N₁₂P₂RhRu₃⋅2CH₂Cl₂ (found): C, 37.85 (37.83); H, 3.61 (3.75); N, 7.16 (7.29).

{Na{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}₂}PF₆.

Ether was allowed to vapor-diffuse into a solution of 20 mg (0.012 mmol) of Cs⊂Co₄Ru₃ and 1.0 mg (0.0061 mmol) of NaPF₆ in 5 ml of 1:1 CHCl₃/MeCN at 0°C. After 72 h, red crystals were collected by decantation and dried under vacuum for 6 h. The crystals were dissolved in 5 ml of CHCl₃ and reprecipitated by allowing Et₂O to vapor-diffuse into the solution at 0°C over the course of 72 h. ¹H NMR (CD₃CN): δ1.681 (s, 90H), 5.485 (s, 30H), and 5.610 (s, 10H).

{Fe{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}₂}(PF₆)₂.

A solution of 13 mg (0.024 mmol) of [Fe(NCMe)₆](PF₆)₂ in 2 ml of MeCN was added dropwise to a solution of 79 mg (0.048 mmol) of Cs⊂Co₄Ru₃ in 12 ml of 1:1 MeCN:CH₂Cl₂. After 24 h, a dark red powder was precipitated by the addition of 30 ml of Et₂O. The powder was collected by filtration, washed with 10-ml portions of THF and Et₂O, and dried under vacuum for 24 h. Yield: 69 mg (79%). ¹H NMR (CD₃CN): δ1.684 (s, 90H), 4.952 (s, 30H), and 5.687 (s, 10H). ESI-MS (m/z): 1678.4. ({Fe{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}₂²⁺). Anal. Calcd for C₁₂₄H₁₃₀Co₈Cs₂F₁₂FeN₂₄P₂Ru₆⋅3CH₂Cl₂ (found): C, 39.10 (39.06); H, 3.51 (3.40); N, 8.62 (8.93).

Crystallography.

Crystals of {Na[Cs⊂Co₄Ru₃]₂}PF₆⋅13MeCN⋅2CH₂Cl₂ were grown by vapor diffusion of Et₂O into a MeCN-CH₂Cl₂ solution (2:1 ratio) of Cs⊂Co₄Ru₃ and NaPF₆. Crystals of {Fe[Cs⊂Co₄Ru₃]₂}(PF₆)₂ grew over the course of 1 week by vapor diffusion of Et₂O into a solution in CH₂Cl₂-MeCN. The crystals were mounted on thin glass fibers with oil (Paratone N, Exxon, Annandale, NJ) before being transferred to a Siemens (Iselin, NJ) Platform/CCD automated diffractometer. Data processing was performed with the integrated program package shelxtl. All structures were solved with direct methods and refined by using full matrix least squares on F² with the program SHELXL-93. Hydrogen atoms were fixed in idealized positions with thermal parameters 1.5× those of the attached carbon atoms. The data were corrected for absorption on the basis of Ψ-scans. Specific details for each crystal are given in Table 1. Full crystallographic details have been deposited with the Cambridge Crystallographic Data Center (nos. CCDC-174664 and -174665). (See Tables 3–12, which are published as supporting information on the PNAS web site, www.pnas.org.)

Table 1.

Crystallographic data for {Na[Cs⊂Co₄Ru₃]₂}PF₆ and {Fe[Cs⊂Co₄Ru₃]₂}(PF₆)₂

Parameter	{Na[Cs⊂Co4Ru3]2}PF13MeCN⋅2CH2Cl2	{Fe[Cs⊂Co4Ru3]2}(PF6)2⋅12MeCN
Chemical formula	C152H173N37Cl4Co8Cs2F6PNaRu6	C148H166N36Co8Cs2F12P2FeRu6
Temp. (K)	153 (2)	193 (2)
Crystal size, mm	0.16 × 0.10 × 0.02	0.30 × 0.24 × 0.06
Space group	P1̄	P21/c
a (Å)	15.330 (4)	22.335 (7)
b (Å)	15.788 (4)	15.936 (5)
c (Å)	21.255 (6)	26.402 (8)
α (deg)	69.696 (4)	90
β (deg)	71.320 (4)	111.558 (5)
γ (deg)	80.682 (5)	90
V (Å3)	4563 (2)	8740 (5)
Z	1	2
Dcalcd (Mg m−3)	1.518	1.573
μ (Mo Kα, mm−1)	1.710	1.815
Minimum and maximum transmission	0.6982/0.9996	0.4527, 0.8942
Reflections measured/independant	33399/8512	65622/64364
Data/restraints/parameters	8512/13/362	64364/0/336
Goodness of fit	0.999	1.017
R1 [I > 2σ] (all data)*	0.1225 (0.2903)	0.1469 (0.3783)
wR2 [I > 2σ] (all data)†	0.2824 (0.3787)	0.1935 (0.4063)
Maximum peak/hole (e-/Å3)	1.329/−1.015	3.506/−3.579

^*R₁ = ∑∥F_o| − |F_c∥/∑|F_o|. 

^† wR₂ = {∑[w(F − F)²]/∑[w(F)²]}^1/2. 

Results and Discussion

{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]_x}^(x−3)+ (x = 3, 4).

The reaction between [K(18-crown-6)][CpCo(CN)₃] and [Cp*Ru(NCMe)₃]⁺ in the presence of CsOTf produces the molecular box {Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₄}⁺, Cs⊂Co₄Ru₄⁺. The structural assignment (Scheme S1) was inferred from its ¹H NMR spectrum, which exhibits one signal each for Cp and Cp* in the appropriate 1:3 intensity ratio. The ESI-MS spectrum (M⁺ at m/z = 1887 atomic mass units) provided evidence for the inclusion of the Cs⁺ at the center of the cage, which is further supported by its ¹³³Cs NMR spectrum (MeCN solution) that shows a single peak at δ1.33 vs. δ34.36 for CsOTf. The IR spectrum features ν_CN bands at 2174 and 2119 cm⁻¹, which are shifted to higher frequencies relative to free [CpCo(CN)₃]⁻ (2119 cm⁻¹), consistent with all cyano ligands serving as bridges. The synthesis of Cs⊂Co₄Ru₄⁺ is modeled after that for the related {[CpCo(CN)₃]₄[Cp*Rh]₄}⁴⁺, except that the formation of Ru-containing cages requires the presence of a templating cation. Attempted synthesis of neutral Co₄Ru₄ cage in the absence of alkali metal salts afforded mainly insoluble, apparently polymeric products.

Scheme 1.

When the synthesis of Cs⊂Co₄Ru₄⁺ is attempted with a deficiency of [Cp*Ru(NCMe)₃]PF₆, one obtains the corresponding Cs⊂Co₄Ru₃ cage, Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃. The violet-colored “defect box” precipitates in excellent yield when [Cp*Ru(NCMe)₃]PF₆ was added to an MeCN solution of CsO₃SCF₃ and [CpCo(CN)₃]₄ in 3:1:4 molar ratio, respectively. In contrast to the ionic cages that we have previously prepared, Cs⊂Co₄Ru₃ is insoluble in MeCN but soluble in Et₂O and THF. Initial characterization of Cs⊂Co₄Ru₃ is based on ESI-MS measurements, with M⁺ at m/z = 1649 and no signals for the box Cs⊂Co₄Ru₄⁺.

The ¹H NMR spectrum of Cs⊂Co₄Ru₃ indicates that it exists as a mixture of two isomers. Each isomer exhibits three singlets in the Cp region and two singlets in the Cp* region with the expected relative intensities of 5:10:5:30:15, respectively. The ¹³³Cs NMR spectrum (THF) also confirms the presence of two isomers characterized by singlets at δ41.46 and δ89.32. The two isomers are proposed to differ in terms of the relative stereochemistry of the Cp and CN_t (terminal) ligands of the cage (see Scheme S1). In contrast, the three CN_t ligands in the cationic bowl Co₄Rh₃²⁺ are oriented toward the exterior of the cage (C_3v symmetry). Because Cp*Ru and Cp*Rh are isosteric, the differing stereochemistry of Co₄Rh₃²⁺ and Cs⊂Co₄Ru₃ must be attributed to the influence of the included (denoted by ⊂) Cs⁺. We propose that the Cs⁺ ion influences the orientation of the CN_t ligands in Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃ via a π-bonding interaction. Preliminary crystallographic analysis (M.L.K., unpublished results) on a weakly diffracting crystal confirms that two CN_t groups are side-bonded to the Cs⁺.

Box-Completion Reactions: Cs⊂Co₄Ru₃ Is a Building Block for Molecular Boxes.

Aside from its novel structure, an exciting aspect of Cs⊂Co₄Ru₃ is its potential as a well defined building block for the assembly of novel molecular cages via Cs⊂Co₄Ru₃ + M → Cs⊂Co₄Ru₃M, a box-completion process. Proof of concept was provided by the finding that Cs⊂Co₄Ru₃ reacts with one equivalent of [Cp*Ru(NCMe)₃]PF₆ to give the completed box Cs⊂Co₄Ru₄⁺ in high yield. No information is presently available on the mechanism by which the CN_t ligands in Cs⊂Co₄Ru₃ rearrange to allow the box-completion process, which results in all three formerly CN_t ligands being related by C₃ symmetry.

The box-completion reaction was extended to the synthesis of the dicationic box {Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃[Cp*Rh]}(PF₆)₂. This synthesis was accomplished by treatment of a solution of Cs⊂Co₄Ru₃ with an equimolar quantity of the dication [Cp*Rh(NCMe)₃](PF₆)₂ (Scheme S1). The transformation occurred at room temperature over the course of 24 h to give Cs⊂Co₄Ru₃Rh²⁺ in 66% isolated yield. The product was characterized by ESI-MS (molecular ion M²⁺ at m/z = 944.4). ¹H NMR spectra showed two Cp signals with relative intensities of 3:1 ratio and two Cp* signals, which also integrate in a 3:1 ratio, a pattern consistent with the expected C_3v symmetry. The ¹³³Cs NMR spectrum (MeCN) of Cs⊂Co₄Ru₃Rh²⁺ exhibits a singlet at δ10.23.

It is instructive to contrast the Cs⊂Co₄Ru₃ + [Cp*Rh(NCMe)₃]²⁺ box-completion reaction with the complementary reaction of Co₄Rh₃²⁺ + [Cp*Ru(NCMe)₃]⁺. The products of these reactions can be identified by ESI-MS by both molecular ions for the cage cations and the ion-paired species containing one or two PF₆⁻ counterions. The Co₄Rh₃²⁺ + [Cp*Ru(NCMe)₃]⁺ reaction afforded a mixture of both Co₄Rh₄⁴⁺ and Co₄Rh₃Ru³⁺, together with trace amounts of Co₄Rh₂Ru₂²⁺. The complexity of the product mixture, especially the formation of substantial amounts of Co₄Rh₄⁴⁺, shows that Co₄Rh₃²⁺ undergoes significant disassembly during the conversion. The difference in the product mixture (vs. that derived from Cs⊂Co₄Ru₃ + [Cp*Rh(NCMe)₃]²⁺) is attributed to the absence of Cs⁺, which in the Co-Ru cages glues together the M₇(CN)₉ framework.

The Double Molecular Boxes [M{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃]}₂]ⁿ⁺.

In the preceding examples, Cs⊂Co₄Ru₃ was deployed as a ligand toward the “half-sandwich” cations Cp*Mⁿ⁺ (M = Rh, Ru), which feature three coordination sites. The use of metal centers with six labile ligands should therefore allow the synthesis of double boxes. Indeed, treatment of solutions of Cs⊂Co₄Ru₃ with NaPF₆ afforded pale red crystals of {Na{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃]}₂}PF₆ (Na[Cs⊂Co₄Ru₃]₂⁺) (Scheme S1). The ¹H NMR spectrum of this salt indicates that the cage has high symmetry, as signals for two types of Cp (relative intensities 1:3) and one type of Cp* are observed.

The structure of {Na[Cs⊂Co₄Ru₃]₂}PF₆, confirmed crystallographically, consists of two {Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃} cages conjoined at an Na⁺ (Fig. 1, Table 2). The Cs⁺ ions are situated at the center of each of the two box-like fragments, similar to other Cs-intercalated boxes (17). The smaller Na⁺ ion is octahedrally coordinated by the six terminal CN ligands from the two defect box fragments. The double cage has idealized D_3d symmetry. The species provides a relatively rare example of an alkali metal ion receptor that contains different alkali metal cations. The Na-N distances are 2.56 Å vs. the Ru-N distances of 2.09 Å; consequently, each Co₄NaRu₃ box is slightly distorted. The extent of this distortion is indicated by the Na^⋅⋅⋅Co edge distance of 5.55 Å vs. the Co^⋅⋅⋅Ru edge distances of 5.12 Å. The Cs⁺-C/N distances occur over the range (3.95(3) − 3.66(2) Å, the broad range due to the fact that Cs⁺ is displaced by 0.27 Å along the Co-Na body diagonal toward the Na. The perspective shown in Fig. 2 shows that the conjoined boxes retain their nearly idealized box-like architectures.

Figure 1.

Molecular structure of the cation {Na{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}₂}⁺ with thermal ellipsoids drawn at the 35% probability level.

Table 2.

Selected bond distances (Å) and angles (°) for the cations in {Na[Cs⊂Co₄Ru₃]₂}PF₆⋅13MeCN⋅2CH₂Cl₂, and {Fe[Cs⊂Co₄Ru₃]₂}(PF₆)₂⋅12MeCN

Parameter	{Na[Cs⊂Co4Ru3]2}+	{Fe[Cs⊂Co4Ru3]2}2+
C(1)-Cs(1)	3.95 (3)	3.657 (10)
C(2)-Cs(1)	3.93 (3)	3.666 (11)
C(3)-Cs(1)	3.89 (3)	3.676 (10)
C(4)-Cs(1)	3.67 (3)	3.620 (11)
C(5)-Cs(1)	3.73 (3)	3.689 (14)
C(6)-Cs(1)	3.73 (3)	3.778 (10)
C(7)-Cs(1)	3.73 (3)	3.661 (13)
C(8)-Cs(1)	3.78 (3)	3.626 (11)
C(9)-Cs(1)	3.73 (3)	3.850 (12)
C(10)-Cs(1)	3.72 (3)	3.624 (10)
C(11)-Cs(1)	3.70 (3)	3.607 (10)
C(12)-Cs(1)	3.76 (3)	3.792 (10)
N(1)-Cs(1)	3.85 (2)	3.588 (8)
N(2)-Cs(1)	3.84 (2)	3.527 (9)
N(2)-Cs(1)	3.86 (2)	3.572 (9)
N(4)-Cs(1)	3.67 (2)	3.554 (8)
N(5)-Cs(1)	3.66 (2)	3.604 (11)
N(6)-Cs(1)	3.66 (2)	3.863 (11)
N(7)-Cs(1)	3.76 (2)	3.594 (11)
N(8)-Cs(1)	3.66 (2)	3.503 (8)
N(9)-Cs(1)	3.61 (3)	3.893 (10)
N(10)-Cs(1)	3.57 (2)	3.551 (8)
N(11)-Cs(1)	3.68 (2)	3.592 (8)
N(12)-Cs(1)	3.64 (2)	3.813 (10)
Na(1)/Fe(1)-N(6)	2.49 (3)	2.207 (8)
Na(1)/Fe(1)-N(12)	2.58 (3)	2.218 (9)
Na(1)/Fe(1)-N(9)	2.65 (3)	2.169 (9)
N(6)-Na(1)/Fe(1)-N(12)	88.6 (8)	87.0 (3)
N(6)-Na(1)/Fe(1)-N(9)	86.7 (7)	88.2 (3)
N(12)-Na(1)/Fe(1)-N(9)	87.4 (8)	88.9 (3)

Figure 2.

View of the M-C-N framework in {Na{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}₂}⁺ with organic ligands removed for clarity.

The formation of double boxes seems to be a general property of Cs⊂Co₄Ru₃ as demonstrated by its reaction with Fe²⁺ sources to give Fe{Cs⊂[CpCo(CN)₃]₄[Cp*Ru]₃}₂(PF₆)₂, {Fe[Cs⊂Co₄Ru₃]₂}(PF₆)₂. Definitive characterization followed from mass spectrometric and crystallographic analyses (Table 1). The geometries of the Na and Fe double boxes are very similar. The Fe double box is more compact, as reflected in the contracted C/N-Cs and Fe⋯N distances, which reflect the smaller radius of Fe²⁺ vs. Na⁺. The corresponding {Co[Cs⊂Co₄Ru₃]₂}(PF₆)₂ was also prepared, further demonstrating the generality of the double-box motif.

Summary

Previous work had established the affinity of alkali metal cations for anionic cages, e.g., {Cs⊂[(CO)₃Mo(CN)₃]₄[Cp*Rh]₄}³⁻ and {Cs⊂[(CO)₃Mo]₆(CN)₉}⁸⁻ (17, 18). The formation of Cs⊂Co₄Ru₄⁺ significantly extends this host-guest behavior to a neutral M₈(CN)₁₂ host for Cs⁺. Even more remarkable is the stability of the corresponding Cs⊂Co₄Ru₃Rh²⁺, which is a Cs⁺ complex of a cationic receptor Co₄Ru₃Rh⁺. Mass spectral data points to the existence of Cs⊂Co₄Ru₂Rh₂³⁺, the Cs⁺ complex of a dicationic host.

The species Cs⊂Co₄Ru₃ represents a Cs⁺ complex of the anionic molecular bowl {[CpCo(CN)₃]₄[Cp*Ru]₃}⁻, which seems to be unstable in the absence of an encapsulated cation. We propose that the bowl is stabilized by Cs⁺⋯CN interactions involving the π bond of cyanide, analogous to the recently observed interaction between Cs⁺ and the triple bond in MeCN (19). Such π-bonding interactions provide the basis for the design of new complexants for Cs⁺, which is relevant to radio-waste remediation (20, 21). The species Cs⊂Co₄Ru₃ is of further interest as a face-capping tridentate ligand, which should allow it to be deployed broadly. Face-capping N₃ ligands have played a significant role in coordination chemistry, and catalysis as exemplified by extensive work on 1,4,7-triazacyclonane (22, 23) and tris(pyrazoyl)borate (24).

The double boxes described in this report are the largest known molecular subunits of a cubic cyanometallate framework. The stability and easy formation of these double boxes suggest the feasibility of synthesizing even larger multicages.

Abbreviations

ESI-MS

electrospray ionization MS

THF

tetrahydrofuran

Cp

cyclopentadienyl

Cp*

pentamethylcyclopentadienyl

PB

Prussian blue

OTf

trifluoromethanesulfonate