Generating and stabilizing Co(I) in a nanocage environment.

A discrete nanocage of core-shell design, in which carboxylic acid groups were tethered to the core and silanol to the shell interior, was found to react with Co2(CO)8 to form and stabilize a Co(I)-CO species. The singular CO stretching band of this new Co species at 1958 cm(-1) and its magnetic susceptibility were consistent with Co(I) compounds. When exposed to O2, it transformed from an EPR inactive to an EPR active species indicative of oxidation of Co(I) to Co(II) with the formation of H2O2. It could be oxidized also by organoazide or water. Its residence in the nanocage interior was confirmed by size selectivity in the oxidation process and the fact that the entrapped Co species could not be accessed by an electrode.

A number of unusual properties of discrete nanocages have emerged since the discovery of carcerand and the recognition of restricted rotational and translational motion for the entrapped molecule.[1,2] By offering an environment in the interior that differs from the surrounding liquid, nanocages have demonstrated the ability to stabilize ionic reaction intermediates[3] and moisture sensitive materials such as phosphorus[4] and alter molecular conformation,[5] reaction regioselectivity,[6,7] reaction rates as well as extent of sequential reaction,[8] and protonation affinity.[9] They have also found applications in targeted drug delivery.[10−12] It is anticipated that new chemical properties would develop as the nanocage design becomes more complex and its functionality more versatile.

Recently, our group has explored creating bifunctional nanocages of a core–shell design, in which the core and the shell interior harbor different functionalities. By linking the core to the shell with a silyl ester bond during synthesis, subsequent hydrolytic cleavage would generate a carboxylic acid on the core and a silanol on the shell. This is a structure that has a hydrophilic interior and a hydrophobic exterior. Thus, it is able to trap and stabilize metal ions in an aprotic solvent.[13] Because of steric crowding, the motion of the limited number of carboxylic acid groups in the structure is restricted, which could create a situation where a metal complex entering the nanocage can access and react with only a small number of carboxylic acid groups and lead to stabilization of unusual metal complexes. Indeed, it was reported that a stable Zn(I) species could be formed in a microporous crystalline silicoaluminophosphate (SAPO-CHA) by reaction of the vapor of a Zn0 compound with an isolated H+ inside a pore,[14] and the resulting Zn-zeolite demonstrated interesting chemical properties such as formation of acetic acid from methane and low-temperature CO oxidation.[15,16] This approach to generate an uncommon oxidation state is atypical. In organometallic complexes, manipulating steric and electronic properties of ligands, such as cyclopentadiene, phosphine, imine, guanidine, and carbene, is the usual method to form complexes with the metal in relatively rare and low formal oxidation states, such as complexes of Cr(I),[17−20] Fe(I),[21,22] Co(I),[23−27] and main group elements.[28]

We report here the formation and stabilization of Co(I) in a discrete molecular, bifunctional, core–shell nanocage by solution chemistry. The nanocage was derived from a nanosphere II, which was formed starting with a malonic acid functionalized spherosilicate core I that was converted to a vinyl-functionalized spherosilicate by reaction with trivinylchlorosilane. Subsequent cross-linking the vinyl groups with 1,4-bis(dimethylsilyl)benzene via hydrosilylation created a porous carbosilane shell (Scheme 1) as previously described.[13] These silyl ester bonds were easily hydrolyzed, e.g., by dispersing 6.7 × 10–3 mmol of II in 10 mL of toluene premixed with a stoichiometric amount of water (0.01 M) and stirring at 100 °C for 24 h. The hydrolysis resulted in carboxylic acid and silanol pairs located inside nanocage III (Scheme 1, inset) that can be used to anchor metal cations in the cage interior. Nearly complete hydrolysis of the silyl ester bond was achieved by using a slight excess of water (∼24 equiv; 1 equiv is 1 mol/mol nanocage). The silanol formed could be detected in 1H NMR, and the carboxylic acid appeared as a small splitting of the νCO stretch in the IR spectrum (Figures S1 and S2). These were the only perceptible spectroscopic changes observed, indicating that the rest of the core–shell structure remained intact. However, the width of the 1H NMR peaks and the overlap of the carbonyl of the silyl ester and carboxylic acid peaks in the IR spectrum prevented quantification of the extent of hydrolysis.

Scheme 1

Spherosilicate Core I with 16 Peripheral Acids from Which Silyl Ester Bonds Were Formed, Nanosphere II Formed by Shell Cross-Linking Hydrosilylation of the Peripheral Vinyl Groups with 1,4-Bis-dimethylsilylbenzene

Inset shows hydrolytic cleavage of silyl ester bonds to form functionalized nanocage III.

Transformation of III into its metalated form Co@III was achieved by reaction with Co2(CO)8 followed by purification to remove excess Co2(CO)8 (Scheme 2). Analysis by inductively coupled plasma indicated that there were ∼4 Co per nanocage at saturation in Co@III (M.W. III ∼ 7900/g). The nature of the encapsulated Co species was characterized by FTIR, cyclic voltammetry, EPR, and magnetization. Changes in the FTIR spectra of a solution of 6.7 × 10–4 M III in dried toluene containing various amounts of Co2(CO)8 are shown in Figure 1. The spectrum for 1 equiv of Co2(CO)8 with III (curve b) showed the absence of the characteristic, intense absorption features of Co2(CO)8 at 2076–2050 cm–1 (terminal carbonyls) or doublet at 1868/1858 cm–1 (bridging carbonyls).[29,30] Instead, a single CO stretch at 1958 cm–1 appeared, together with a broad carboxylate band at ∼1589 cm–1 that overlapped with the slightly diminished carbonyl stretch of the carboxylic acid of III. After adding a total of 12 equiv of Co2(CO)8, the carboxylate band intensity increased further, suggesting that each nanocage reacted with more than one Co2(CO)8 (curve c). Peaks characteristic of unreacted Co2(CO)8 were also present. After removing the unreacted Co2(CO)8, the IR spectrum of the resulting clear orange solution of Co@III (curve d) displayed a sharp 1958 cm–1 carbonyl peak, peaks at 1716 cm–1 (carboxylic acid) and 1594 cm–1 (carboxylate), and those associated with the rest of the nanocage. It is worthy to note that, without hydrolyzing the silyl ester bond, the nanosphere II did not react with Co2(CO)8, as monitored by FTIR (Figure S4). In the literature, the carbonyl peak at 1958 cm–1 has been assigned to a Co(I) species.[25,31,32]

Scheme 2

Preparative-Scale Reaction and Purification To Form Co@III

Figure 1

IR spectra of (a) III, (b) III + 1 equiv of Co2(CO)8 after 34 h at rt; (c) III + 12 equiv of Co2(CO)8, (d) Co@III which is sample c after removal of unreacted Co2(CO)8 with Smopex-301 at 50 °C in 4 h.

The only change in the 1H NMR spectrum after reacting III with Co2(CO)8 was the splitting of the broad, unresolved peak of the phenyl proton of the cage wall (7.3–7.7 ppm) (Figure S5), suggesting interaction of the entrapped Co with the carbosilane shell and their close proximity. Cyclic voltammetry of a THF solution of Co@III was used to determine whether the Co species were located inside the cage or on the outside wall. The CV curve of Co2(CO)8 in THF showed redox peaks between −2 and 1 V vs SCE from the Co1+/Co0 couple,[33] together with those of the internal standard ferrocene at 0.25–0.75 V. The redox peaks for Co were absent for the solution of Co@III, indicating that the Co species were inside the cage and not accessible to the electrode (Figure S6).

The number of CO ligands per Co was determined by thermally decomposing the Co species in Co@III at 80 °C and quantifying the CO released. The average stoichiometry was 1 CO per Co (Supporting Information (SI) section VI). The decomposition caused the disappearance of the 1958 cm–1 carbonyl peak.

The EPR spectra of Co@III before and after exposure to O2 (Figure 2) supports the assignment of an initial state of Co(I). The spectrum of Co@III showed a small signal at g = 4.0 from high-spin (S = 3/2) Co(II). Quantitation indicated that this sample contained ∼0.1 mM of Co(II), accounting for ∼10% of the total Co in the sample as prepared, which can be attributed to oxidation during sample preparation. After exposure to O2 for 90 min the solution changed color from orange to blue-purple. The EPR spectrum showed a corresponding increase in the signal intensity at g = 4.0, the signal now accounting for 70% of the Co in the sample. This oxidation of Co(I) to Co(II) was accompanied by the disappearance of the 1958 cm–1 band in the IR, although the rest of the spectrum remained unchanged, indicating that the nanocage structure was stable to oxidation (Figure S12). This oxidized species is labeled Co(ox)@III (MW = 8248 g/mol).

Figure 2

EPR spectra of (A) Co@III and (B) Co@III after 90 min exposure to O2. Spectral conditions: microwave frequency = 9.36 GHz, microwave attenuation = 20 dB, modulation amplitude = 13.2 G, modulation frequency = 100 kHz, T = 7 K.

The magnetic susceptibility of Co@III and Co(ox)@III were measured in Ar over the range 5–300 K (Figure 3). Both samples were paramagnetic, and the magnetization followed the Curie–Weiss law.[34] From the molar magnetic susceptibility, and assuming that all cobalt centers are independent and homogeneously dispersed in the nanocage, a room temperature effective magnetic moment of 3.55 μB was calculated for Co@III, which exceeded the expected spin-only value of 2.83 μB for a high-spin d8 (S = 1) system, but was comparable to reported values of other cobalt(I) complexes.[25,35] Such a difference has been attributed to a substantial orbital angular momentum contribution from the cobalt ion. The Co(II) species in Co(ox)@III was in a 3d7 high-spin (S = 3/2) state, and its room temperature effective magnetic moment of 5.0 μB was within the range observed for high-spin cobalt(II) complexes with pseudotetrahedral or lower symmetry and suggestive of a quartet magnetic ground state.[25,36]

Figure 3

Temperature dependence of mass magnetic susceptibility of Co@III (black squares) and Co(ox)@III (red dots). The lines are fitting curves.

The UV–vis spectrum of Co@III in toluene showed a very intense metal-to-ligand charge transfer band at 396 nm (ε = 6.27 × 105 M–1·cm–1), whereas Co2(CO)8 showed absorption at 350 and 280 nm (Figure S10a and c). Upon exposure to O2, the absorbance at 396 nm decreased with a pseudo-first-order rate constant of 0.014 min–1 (Figure S10b), and two weak broad bands at 530 nm (ε = 1045 M–1·cm–1) and 589 nm (ε = 1134 M–1·cm–1) appeared that were indicative of d–d transitions of high-spin Co(II) (Figure S10c). These spectroscopic changes were visible, as the solution changed from orange to pale blue-purple, corresponding to that seen upon oxygenation of the concentrated EPR sample, as described above.

The product of the reaction of Co@III with O2 was extracted with water and analyzed for H2O2 by titration with potassium titanium oxalate (see SI). For a reaction that started with 4 × 10–4 mmol of Co, 1.18 × 10–4 mmol of H2O2 was detected in the aqueous phase, which corresponded to a 59% yield. Trace amounts of benzyl aldehyde and benzyl alcohol were detected by GC/MS in the toluene phase. With handling losses considered, the reduction of O2 by Co@III and hydrolysis to form H2O2 appeared to be nearly quantitative.

Co@III can be oxidized by organoazide also, and the size selectivity of III can be demonstrated by using different organoazides. Two azides of different molecular sizes, phenyl and adamantyl, were tested by adding 1 equiv of oxidant to Co@III in toluene (6.27 × 10–4 M) and following the reaction with UV spectroscopy (Figure S17). In both cases, the orange solution became pale yellow at the end of the reaction, and the 396 nm band disappeared. The change of color was complete within 2 h for phenyl azide, but took about 12 h for adamantyl azide. Thus, the smaller phenyl azide entered the nanocage much faster than adamantyl azide and further supported the interior location of Co. For both reactions, EPR analyses of the product solution could not detect any signal, and the high reactivity of the nanocage prevented useful analysis by mass spectroscopy. The results from UV–vis spectroscopy and EPR, and the fact that the color remained pale yellow after exposure to air, suggested that the Co(I) in Co@III was oxidized to a Co(III) species. At present, it is not known whether the product is a Co-imido species or Co coordinated to more than one organoazide. In the literature, there are examples of organoazide oxidation of Co(I) to cobalt(III) imido complexes.[35,37,38] H2O was also effective in oxidizing Co@III.

In order to confirm the relevance of the nanocage environment in this process, the reaction of Co2(CO)8 with the spherosilicate template I in toluene was examined. I possesses 16 carboxylic acid groups on the periphery. This reaction resulted in a pink precipitate of rather uniform, spherical globules about 200 nm in diameter (Figure S13). XPS of the precipitate indicated formation of Co(II) (Figure S16). Apparently, Co2(CO)8 was oxidized by reaction with the carboxylic acids to form a network of spherosilicate carboxylate cross-linked with Co(II).

Thus, we have demonstrated that a nanocage with a core–shell structure, containing a prescribed number of common functional groups and restricted space for movement of complexes, can be used to generate and stabilize metal complexes of an uncommon oxidation state with a high degree of uniformity. In this example, this is achieved by limiting the number of carboxylic acid groups accessible to each Co carbonyl complex. The concept is general, and the approach has the potential to greatly expand the tools for stabilizing low-valent metal ions, in addition to the standard approach of manipulating the properties of ligands that include Cp, phosphines, and others. When coupled with the properties of hydrophilic interior and hydrophobic exterior, these nanocages can be used with other metal complexes of a variety of properties, creating opportunities to explore new metal complexes of uncommon oxidation states.